- mvu_vvu_axi.sv: USE_COMPRESSOR gating, MAX_IN_FLIGHT safety floor
- add_multi.sv: CATCH_COMP macro for lane reduction compressors
- mvu_vvu_axi_wrapper.v: COMP_PIPELINE_DEPTH parameter propagation

- Add compressor eligibility checks (_is_dotp_comp_eligible, etc.)
- Call generate_dotp_comp() and generate_add_multi_comps()
- Add generated files to RTL file list
- Propagate COMP_PIPELINE_DEPTH and USE_COMPRESSOR template vars

- Core compressor unit tests (run_tests.sh): 21 XSim-based configs
- dotp_comp integration tests: 8 configs (Versal + 7-Series)
- add_multi integration tests: 8 configs (DSP lane reduction)
- MVU synthesis tests for timing closure validation
- REPORT.md: Status, known issues, benchmarking results
- TEST_GUIDE.md: Testing procedures and troubleshooting

Correct LUT6_2 predicate order and CARRY4 carry-in wiring for
MuxCYRippleSum and MuxCYPredAdder counters on 7-Series FPGAs.

Key fixes:
- Swap O5/O6 outputs to match hardware behavior
- Add initial carry-in=0 for first LUT in chain
- Implement MuxCYPredAdder for horizontal multi-column absorption

Re-enables gate absorption optimization for 7-Series targets.

Compressor path handles full weight range (no narrow weight restriction).
DSP path uses NARROW_WEIGHTS module parameter for range adjustment.

This enables RTL MVAU on 7-Series (DSP48E1) with full weight ranges.

Add COMP_PIPELINE_DEPTH and USE_COMPRESSOR template vars to VVU
for consistency with MVU. VVU does not use compressor (different
compute pattern), so USE_COMPRESSOR=0.

Aggregates CATCH_COMP entries from all MVAU nodes into a single
add_multi.sv file during out-of-context synthesis. Required for
add_multi compressor path to work in full builds.

Ignore Claude Code workspace files and benchmark results directories.

Remove test scripts and templates that are redundant with pytest coverage:
- run_add_multi_comp_tests.sh + add_multi_comp_tb_template.sv/.tcl
- run_mvu_add_multi_comp_tests.sh + mvu_add_multi_comp_tb_template.sv/.tcl
- run_mvu_comp_synth_tests.sh + mvu_comp_synth_tb_template.tcl

These remain available in the compressor-benchmarking branch for
detailed performance analysis. Core validation covered by:
- run_tests.sh (compressor unit tests)
- run_dotp_comp_tests.sh (dotp_comp integration)
- run_mvu_comp_tests.sh (full MVU integration)
- pytest test_fpgadataflow_rtl_mvau (FINN node tests)