CompatHelper: bump compat for AbstractPPL to 0.15 for package benchmarks, (keep existing compat)

[github-actions]

This pull request changes the compat entry for the AbstractPPL package from 0.14 to 0.14, 0.15 for package benchmarks.
This keeps the compat entries for earlier versions.

Note: I have not tested your package with this new compat entry.
It is your responsibility to make sure that your package tests pass before you merge this pull request.

[github-actions]

DynamicPPL.jl documentation for PR #1401 is available at:
https://TuringLang.github.io/DynamicPPL.jl/previews/PR1401/

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.30%. Comparing base (d2052a1) to head (ab519aa).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1401   +/-   ##
=======================================
  Coverage   82.30%   82.30%           
=======================================
  Files          50       50           
  Lines        3543     3543           
=======================================
  Hits         2916     2916           
  Misses        627      627           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

Benchmarks @ ab519aa

Performance

Performance Ratio:
Ratio of time to compute gradient and time to compute log-density.
Warning: results are very approximate! See benchmark notes for more context.

===================================================================================================
                                               eval                       gradient                 
                                            ----------  -------------------------------------------
Model                        dim    linked      primal     FwdDiff    RvsDiff    Mooncake    Enzyme
---------------------------------------------------------------------------------------------------
Simple assume observe*         1     false     5.87 ns       10.22    1221.27       28.98      6.22
Simple assume observe*         1      true     24.2 ns        2.50     318.59        7.40      1.52
Smorgasbord                  201     false     6.45 μs       74.57     124.39        6.21      8.65
Smorgasbord                  201      true     8.81 μs       64.47     127.93        5.26      5.90
Loop univariate 1k          1000     false     19.4 μs     1204.96     280.46        7.31      6.01
Loop univariate 1k          1000      true     20.7 μs     1567.59     266.19        6.96      5.65
Multivariate 1k             1000     false     22.3 μs      380.33      81.43        9.93      2.71
Multivariate 1k             1000      true     19.8 μs      267.40      66.79       11.21      2.76
Loop univariate 10k        10000     false    188.0 μs    14480.71     316.00        7.80      6.11
Loop univariate 10k        10000      true    204.0 μs    14666.25     302.36        7.19      5.68
Multivariate 10k           10000     false    203.0 μs     5686.41      95.69       11.49      2.18
Multivariate 10k           10000      true    203.0 μs     5978.88      94.36       11.22      2.17
Dynamic                       15     false     1.38 μs         err      42.47       15.00     11.13
Dynamic                       10      true     1.91 μs        1.96      60.04       18.95     20.75
Submodel*                      1     false     5.88 ns       10.26    1238.91       28.99      6.28
Submodel*                      1      true     5.88 ns       10.30    1451.62       29.19      6.34
LDA                           12      true     24.0 μs        0.45       1.98       31.92       err
===================================================================================================

Rows marked * have t(logdensity) below about 100 ns; their ratios can be dominated by timer floor, fixed overhead, and run-to-run variation. For those rows, raw t(grad) is more meaningful than t(grad)/t(logdensity).

Main @ d2052a1

===================================================================================================
                                               eval                       gradient                 
                                            ----------  -------------------------------------------
Model                        dim    linked      primal     FwdDiff    RvsDiff    Mooncake    Enzyme
---------------------------------------------------------------------------------------------------
Simple assume observe*         1     false     5.87 ns       10.26    1102.26       28.98      6.75
Simple assume observe*         1      true     24.2 ns        2.48     321.83        7.00      1.62
Smorgasbord                  201     false      6.5 μs       65.56     127.06        6.23      8.87
Smorgasbord                  201      true     8.77 μs       63.80     123.86        5.21      5.95
Loop univariate 1k          1000     false     19.3 μs      974.65     277.98        7.92      6.05
Loop univariate 1k          1000      true     20.8 μs     1368.63     262.60        7.03      5.55
Multivariate 1k             1000     false     22.7 μs      339.67      76.55        9.55      2.57
Multivariate 1k             1000      true     25.7 μs      269.75      59.93        8.81      2.49
Loop univariate 10k        10000     false    200.0 μs    10432.89     283.39        7.69      5.76
Loop univariate 10k        10000      true    205.0 μs    10923.00     285.41        6.77      6.79
Multivariate 10k           10000     false    202.0 μs     5139.87      82.46       11.06      2.14
Multivariate 10k           10000      true    201.0 μs     5213.51      88.29       11.08      2.14
Dynamic                       15     false     1.38 μs         err      42.87       15.17     11.15
Dynamic                       10      true     1.96 μs        1.91      56.78       21.07     19.11
Submodel*                      1     false     5.87 ns       10.25    1354.38       29.04      6.60
Submodel*                      1      true     5.87 ns       10.24    1457.19       29.55      6.60
LDA                           12      true     22.5 μs        0.44       2.04       33.87       err
===================================================================================================

Environment

Julia Version 1.11.9
Commit 53a02c0720c (2026-02-06 00:27 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: Linux (x86_64-linux-gnu)
  CPU: 4 × AMD EPYC 7763 64-Core Processor
  WORD_SIZE: 64
  LLVM: libLLVM-16.0.6 (ORCJIT, znver3)
Threads: 1 default, 0 interactive, 1 GC (on 4 virtual cores)