* perf(terminal): eliminate dense control and frame gate regressions
* test(terminal): keep gate labels in valid expect shape
* test(terminal): expose the surviving sub-threshold control-density case
The only adverse strip fixture sat at 50% control density, which is exactly
where the fallback fires and wins. A shape at 31 controls per 64-unit block
evades the trigger and still loses to the per-character legacy (0.67x), so the
benchmark structurally could not show it.
Add that fixture, pin both density literals in the staleness guard so a retune
fails loudly instead of silently measuring a boundary that moved, and export
the probe constant the equivalence test was hardcoding.
* perf(terminal): scan output frames by code unit, not per code point
iterateTerminalOutputFrameChunks walked `for (const part of data)`, materializing
a 1-2 character string per code point and calling terminalStreamByteLength on
each, while accumulating the frame text with `chunk += part`.
The accumulator was never needed: `chunk` only ever reconstructs the contiguous
substring data[chunkStart..end), and `startSeq + chunkStartOffset + chunk.length`
collapses algebraically to `startSeq + end`. Track two integer offsets and emit
data.slice(chunkStart, end) instead, computing UTF-8 width inline from charCodeAt.
Also short-circuits the cap gate on UTF-16 length before measuring UTF-8 bytes,
which is sound because UTF-8 length is never below UTF-16 length.
2.4-6.1x across payload shapes, stable across reruns. This runs per terminal
output batch and per snapshot chunk.
Extracted to its own module along a real seam (the chunk-emission concern plus
its two types and cap gate); methods/terminal.ts shrinks by 85 lines. No
max-lines suppression added and the baseline is untouched.
Co-authored-by: Orca <help@stably.ai>
* fix(terminal): preserve chunk sequence rounding
Co-authored-by: Orca <help@stably.ai>
---------
Co-authored-by: Orca <help@stably.ai>
Neil
