Give the KAOS bridge the same E-mode treatment, and keep it current

[solutionphil]

Verifying the alpha update on the dev printer turned up something the AddOn.cfg fix cannot reach: KAOS declares [gcode_macro PHROZEN_TOOLCHANGE] a second time in kaos-ams-bridge.cfg, and a later declaration wins per option. So with KAOS on, the macro that actually runs is that one — and it had the defect twice over.

The AMS branch was worse than the kit's

Its own comment says M83 goes first to close "the kit's standing G1 E-1 before M83 latent defect on this path". It does — but it never hands the mode back. A slicer running absolute E would have every extrusion after the first tool change reinterpreted as relative, for the rest of the print. That is a bigger fault than the one it fixes.

Confirmed by grep: exactly one M83 in the whole bridge, and nothing anywhere restores the mode — not ORCA_PURGE, not PRZ_SPITTING_END.

The no-AMS branch still had the original defect

Its comment says why: "byte-identical to stock". It was, when written. The kit has since fixed that line, so the copy was stale and the claim no longer true.

Both branches now read the mode once at the top and hand it back. Rendered against Klipper's own jinja2, all four combinations:

ams	mode on entry	renders
1	M82	M83 → retract → M82
1	M83	M83 → retract
0	M82	M83 → retract → M82
0	M83	M83 → retract

It had to survive KAOS_ON / KAOS_OFF

Our bridge configs are copied into printer_data/config when KAOS is activated and never again. $SELF_DIR/config is inside the kit clone, so a kit update refreshes the source and leaves the copies — a printer that has had KAOS on since before an update keeps running the old ones. The same "the correction never reaches the printer" shape as the rest of this week.

So kaos-guard.sh, which already reconciles the spit patch before every klipper start, now refreshes our bridge configs too — keyed on the include line actually present in printer.cfg, exactly like sync_spit_patch. A file whose include is gone is never put back, so KAOS_OFF stays off. KAOS's own files come from the upstream cache and are untouched.

Measured against a copy of the dev printer's config with an isolated KAOS_HOME:

Case	Result
include present, bridge stale	refreshed, and says so
second run	silent
include removed, file deleted	not put back
KAOS off	the AddOn.cfg macro runs — already current through #@REVISE