[egg-rolls]

Summary

This PR fixes several issues found during code review of the SuperRTL codebase.

1. Fix case statement depth counting in latch detection

• File: src/superrtl/tools/review.py
• Issue: The _check_latch_inference function initialized depth = 0 when detecting a case statement, but the first case keyword was then matched by "case" in case_line inside the loop, incrementing depth to 1 immediately. This meant the loop would look for one extra endcase that didn't exist, potentially running past the actual end of the case block.
• Fix: Changed depth = 0 to depth = 1 and added j != i guard to skip double-counting the first case.

2. Improve camelCase naming detection

• File: src/superrtl/tools/review.py
• Issue: The _check_naming function excluded all signals starting with clk or rst from camelCase warnings, including longer names like clkDivider or rstGenerator that should use snake_case.
• Fix: Added length check to only allow short common prefixes (6 chars or less) like clkDiv, rstN.

3. Fix I2C 10-bit address decoding

• File: src/superrtl/tools/protocol_decode.py
• Issue: The address_bits config parameter was read from config but never used in the actual decoding logic. All addresses were decoded as 7-bit regardless of the setting.
• Fix: Implemented proper 10-bit I2C address parsing per the I2C specification (first byte = 11110xx0, second byte = full address bits).

4. Improve signal name matching precision

• File: src/superrtl/tools/protocol_decode.py
• Issue: The _get_signal_values function used simple substring matching (in), which could incorrectly match signals like tx_data when looking for tx, or spi_miso when looking for sck (because sck appears in spi_sck).
• Fix: Implemented prioritized matching: exact match > prefix match (with delimiter boundary check for _, ., [) > substring fallback.

5. Enable I2C ACK detection

• File: src/superrtl/tools/protocol_decode.py
• Issue: I2C ACK detection code was present but commented out with pass.
• Fix: Enabled the ACK detection logic to read SDA state during the 9th clock cycle.

Testing

• All 154 tests pass (8 skipped due to missing EDA tools)
• No new test failures introduced