CompatHelper: bump compat for AbstractPPL to 0.15 for package test, (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 test.
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 #1400 is available at:
https://TuringLang.github.io/DynamicPPL.jl/previews/PR1400/

[github-actions]

Benchmarks @ d1d3e02

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       10.28    1215.00       29.04      6.30
Simple assume observe*         1      true     24.2 ns        2.50     320.16        7.03      1.54
Smorgasbord                  201     false     6.44 μs       66.81     123.45        6.36      8.81
Smorgasbord                  201      true     8.81 μs       63.40     130.70        5.28      6.02
Loop univariate 1k          1000     false     19.6 μs      901.78     276.54        7.39      5.93
Loop univariate 1k          1000      true     20.8 μs     1355.04     263.02        6.87      5.68
Multivariate 1k             1000     false     23.6 μs      334.93      73.40        8.62      2.79
Multivariate 1k             1000      true     21.9 μs      286.73      64.83       10.37      2.66
Loop univariate 10k        10000     false    191.0 μs    11631.56     301.56        7.68      6.13
Loop univariate 10k        10000      true    206.0 μs    11371.78     292.93        7.13      5.68
Multivariate 10k           10000     false    206.0 μs     4942.57      86.09       11.21      2.17
Multivariate 10k           10000      true    205.0 μs     4834.85      86.32       11.15      2.14
Dynamic                       15     false     1.41 μs         err      44.29       14.43     11.64
Dynamic                       10      true      2.0 μs        2.05      57.84       20.62     18.75
Submodel*                      1     false     5.88 ns       10.46    1397.23       29.25      6.30
Submodel*                      1      true     5.88 ns       10.34    1466.89       29.27      6.36
LDA                           12      true     25.2 μs        0.48       2.02       30.10       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      5.2 ns       10.59    1252.87       30.92      6.85
Simple assume observe*         1      true     18.9 ns        2.96     419.91        8.20      1.79
Smorgasbord                  201     false     9.53 μs       51.29      84.05        7.25      6.14
Smorgasbord                  201      true     19.7 μs       28.83      51.98        3.78      2.81
Loop univariate 1k          1000     false     49.8 μs      408.40     106.00        3.38      2.61
Loop univariate 1k          1000      true     49.6 μs      556.14     110.66        3.56      2.65
Multivariate 1k             1000     false     38.4 μs      233.95      43.38        4.42      1.49
Multivariate 1k             1000      true     38.4 μs      240.73      41.16        5.62      1.73
Loop univariate 10k        10000     false    203.0 μs    12534.01     273.38        6.94      6.24
Loop univariate 10k        10000      true    209.0 μs    13363.70     271.30        6.51      5.92
Multivariate 10k           10000     false    237.0 μs     6993.96      72.59        8.89      1.86
Multivariate 10k           10000      true    236.0 μs     6880.57      72.44        9.14      1.84
Dynamic                       15     false     2.31 μs         err      31.73       12.07      8.66
Dynamic                       10      true     3.01 μs        2.08      40.50       10.03     14.20
Submodel*                      1     false      5.2 ns       21.26    1709.00       59.09     12.25
Submodel*                      1      true      5.2 ns       21.60    1794.86       58.21     12.27
LDA                           12      true     27.7 μs        0.62       2.07       25.90       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)