fix: priority 1-4 debug pass — retry loop, SSE timeout, rocprof CSV parser, silent failures

fix: remove convolution_2d from evidence table, clarify docstring

feat: add real rocprof CSV evidence and sync all benchmark docs (MI300X gfx942, ROCm 7.0, May 8 2026)

feat: replace demo_artifact with real MI300X measurements (ROCm 7.0, May 2026)

render fix req

docs fix

hipify fix

fixing bugs23

fixing docs proof

fix: ROCm compile flags for gfx942 MI300X - verified on AMD DevCloud

v2

fix render

fixing bugs

fixv2

Initial commit