Batch proving with discriminant reuse (stacked on #325)

[hoffmang9]

Summary

• add the Trick 2 batch-proving path from the bbr fork (f3c73bf, 750df66) so jobs sharing (challenge, size_bits, x0) can reuse one squaring trajectory
• include the follow-up RSS stabilization commit (445cb0d) for the batch finalization path
• stack this branch on top of PR Add streaming one-wesolowski compaction APIs #325 so implementation work can start before Add streaming one-wesolowski compaction APIs #325 merges; once Add streaming one-wesolowski compaction APIs #325 merges, this PR will be updated/rebased to present only PR2 delta

Notes

• this is intentionally opened as a draft stacked PR before Add streaming one-wesolowski compaction APIs #325 is merged
• current diff includes Add streaming one-wesolowski compaction APIs #325 commits because of the stacked base

Test plan

• wait for Add streaming one-wesolowski compaction APIs #325 merge, then rebase/update this branch onto main
• run full CI matrix after rebase
• run fast-wrapper batch API validation in downstream integration

Made with Cursor

---

Note

High Risk
Adds a large new C/C++ proving path that changes how Wesolowski proofs are generated and introduces new multi-threaded batch finalization and fast-path counter slot allocation, which could affect correctness and stability under concurrency.

Overview
Adds a new fast_wrapper C ABI for compaction-oriented one-Wesolowski proving that can stream bucket updates during squaring (requires known y_ref), optionally using an incremental GetBlock mapping to avoid expensive per-block exponentiation, plus a batch API that reuses a single squaring trajectory across many jobs sharing (challenge, x, discriminant_size_bits) and offloads per-job finalization to a small worker pool.

Updates the core squaring loop integration to support batch lifecycle hooks (OnBatchStart/OnBatchReplay), introduces quiet_mode to suppress stdout in library use, and assigns per-thread fast-counter slots via vdf_fast_pairindex() to reduce collisions when running multiple VDFs in-process.

Build/CI updates ensure cmake is present on macOS runners and extend Makefile.vdf-client with optional PIC builds and a new libchiavdf_fastc.a target; adds docs describing the new bluebox compaction optimizations.

^{Reviewed by Cursor Bugbot for commit 8e76d33. Bugbot is set up for automated code reviews on this repo. Configure here.}

[hoffmang9]

@Ealrann Assuming these two PRs are merged - is there anything else you need to be able to abandon your chiavdf fork and just use a release of this repo?

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 88602ba. Configure here.}