chore: circuit to polys cleanup

[ledwards2225]

Two cleanups in trace-to-polynomials land. Originally motivated by small perf wins but is attractive for the code simplification:

1. Simpler permutation polynomial construction. The old code built a large intermediate table and walked it to produce the sigma/id polynomials. The new code
writes those polynomials directly in three steps: identity init, cycle linkages, public-input update.

2. Each block has only the non-gate selectors plus (at most) one gate selector. Previously every block declared all eight or nine possible gate selectors using the hacky ZeroSelector. ZeroSelector class is gone plus lots of simplification in the block classes.

[ledwards2225]

Bench: realistic Chonk flow, dedicated remote (HARDWARE_CONCURRENCY=16)

Flow: schnorr+deploy_tokenContract_with_registration+sponsored_fpc, merge-train pinned inputs 341842bf. Single run on the isolated EC2 box.

Region	before (ms)	after (ms)	Δ
ChonkAPI::prove (total)	6711.6	6643.1	−1.0%
ChonkAccumulate	4345.6	4291.9	−1.2%
construct_trace_data	138.5	133.0	−4.0%
compute_permutation_argument_polynomials	56.8	45.2	−20.4%
populate_wires_and_emit_cycles	21.8	19.3	−11.5%
populate_selectors	16.3	14.6	−10.6%
fill_copy_cycles	19.0	17.1	−9.7%

The regions the PR directly touches all show double-digit improvements: dropping the intermediate PermutationMapping struct buys −20% on compute_permutation_argument_polynomials, and de-virtualising the selector reads/writes (single concrete Selector<FF> + kind tag instead of a Selector / ZeroSelector / SlabVectorSelector hierarchy) buys ~10% across the trace-construction loop. Total proving wall-clock is −1%. Sub-percent moves elsewhere are within single-run noise on this machine.

[ledwards2225]

the condition this test is checking is now structurally impossible to deviate from

[ledwards2225]

ditto

[ledwards2225]

massive code simplification here primarily due to constructing the permutation polys directly rather than passing through the intermediate Mapping/PermutationMapping. also a perf improvement since it avoids many small allocations