fix(get_logs): paginate through full block range on max results exceeded

[gakonst]

Summary

Fixes the rpc-tester failing every nightly run due to get_logs exceeding the 20K result limit.

Motivation

The rpc-tester has been failing on every single nightly run — including when re-execute succeeds. The failure is always the same:

24432145..=24432176 ❌
    get_logs: ❌ rpc1: ErrorResp(ErrorPayload { code: -32602, message: "query exceeds max results 20000, retry with the range 24432145-24432171", data: None })

All per-block tests pass (32/32 ✅). The failure is only in test_block_range's unfiltered get_logs across 32 blocks, which produces >20K logs on mainnet.

The previous get_logs_with_retry had two bugs:

1. It only retried once with the suggested narrower range — which could also exceed the limit
2. In the comparison context (test_rpc_call), the retry only narrowed one provider's range while the other returned the full range, making comparison invalid

Changes

• Replaced single-retry with recursive pagination in get_logs_with_retry (crates/rpc-tester/src/get_logs.rs)
• Parses the RPC's suggested range to determine chunk size, splits the original range into chunks, and recursively fetches each chunk
• Adds overflow protection, non-progress guards, and range validation
• Both providers now independently paginate to return complete results for the full range

Testing

cargo clippy --workspace --all-targets -- -D warnings  # clean
cargo test -p rpc-tester                                # 5/5 pass

Thread: https://tempoxyz.slack.com/archives/C0A87C21805/p1770821518221609