chore: set precomp window size by platform

[abelmega]

Summary

This PR centralizes the default precomputation window size used by Committer initialization and makes it platform-aware.

The main change is:

• use 3 on zkvm + riscv32
• keep 11 on all other platforms

Changes

• add banderwagon::platform::DEFAULT_PRECOMP_WINDOW_SIZE
• export platform from banderwagon
• update salt::trie to consume the shared platform default instead of keeping a local constant

Motivation

PRECOMP_WINDOW_SIZE is a platform tuning parameter rather than trie-local logic. Moving it into banderwagon makes the ownership clearer and avoids duplicating platform-specific configuration in salt.

This also allows zkVM to use a smaller window size without changing the default behavior on non-zkVM platforms.

[github-actions]

Performance Benchmark Comparison

Compared 5 benchmark(s) against the latest main baseline.

Detailed Comparison

Benchmark	Baseline Throughput (Kelem/s)	New Throughput (Kelem/s)	Change
update 10000 KVs/1 threads	76.21	78.64	+3.18%
update 10000 KVs/2 threads	142.73	143.67	+0.66%
update 10000 KVs/4 threads	264.68	268.40	+1.41%
update 10000 KVs/8 threads	450.43	432.70	-3.94%
update 10000 KVs/16 threads	612.24	596.93	-2.50%

[flyq]

Other call sites still hardcode 11

After this PR, Committer::new(..., 11) still appears at:

• ipa-multipoint/src/main.rs:14 — benchmark binary
• banderwagon/src/salt_committer.rs:55 — doc comment
• banderwagon/src/salt_committer.rs:519,561 — tests

The test/benchmark sites are defensible (they're measuring fixed parameters), but if the centralization story matters, at least update the doc comment in salt_committer.rs:55 to mention platform::DEFAULT_PRECOMP_WINDOW_SIZE. The ipa-multipoint benchmark is your call — leave it pinned to 11 for stable measurements, or migrate it for consistency.

[abelmega]

Other call sites still hardcode 11

After this PR, Committer::new(..., 11) still appears at:

• ipa-multipoint/src/main.rs:14 — benchmark binary
• banderwagon/src/salt_committer.rs:55 — doc comment
• banderwagon/src/salt_committer.rs:519,561 — tests

The test/benchmark sites are defensible (they're measuring fixed parameters), but if the centralization story matters, at least update the doc comment in salt_committer.rs:55 to mention platform::DEFAULT_PRECOMP_WINDOW_SIZE. The ipa-multipoint benchmark is your call — leave it pinned to 11 for stable measurements, or migrate it for consistency.

Updated the remaining Committer::new(..., 11) call sites to use platform::DEFAULT_PRECOMP_WINDOW_SIZE as well.

This includes:

• the salt_committer doc example
• the fixed-parameter tests in banderwagon
• the ipa-multipoint benchmark binary

I also extended the existing correctness sweep to cover window_size = 3.