Generative-Logic/GL: Performance update for Peano

• feat(core): rewrite prover in C++17 (MSVC 2022), batched CE filtering, unify to single "full run" mode — 5h → 7s and 4 GB → 1 GB peak.

Summary

Prover rewritten in modern C++17 and built with MSVC 2022. Execution is parallelized across all logical cores on a single machine.

Counterexample (CE) filtering added.

Runtime and memory improved substantially.

Measured impact (Dell G16 7630, release x64)

Prover runtime: ~5 hours → ~7 seconds.

Peak memory (RSS): ≈ 1 GB during CE filtering.

User-visible changes

Only one run mode remains: full run. (Quick/almost full removed.)

Default execution uses all logical cores; CE filter batch size scales with core count.