test: reduce mutation observer delay from 200ms to 0ms

[pskelin]

Purpose

Experimental PR to measure the impact of the mutation observer delay on CI test execution speed.

Change

In packages/base/src/Render.ts, the mutation observer timer in scheduleRenderTask currently waits 200ms before resolving the render task promise. This PR reduces that delay to 0ms to evaluate whether tests run faster on CI.

-					}, 200);
+					}, 0);

Note

Not intended for merge — purely a benchmark to compare CI run duration against the baseline.

[sap-ui5-webcomponents-release]

🚀 Deployed on https://pr-13642--ui5-webcomponents-preview.netlify.app

[pskelin]

Finding: no measurable speedup

Compared the Test step durations on the same passing jobs in this PR (0ms) vs baseline run #26949481414 on main (8247700b, 200ms):

Job	PR #13642 (0ms)	main baseline	Δ
check (base)	77s	78s	−1s
check (compat)	39s	43s	−4s
check (fiori)	196s	181s	+15s
check (main:cy:suite-1)	238s	224s	+14s
check (main:cy:suite-3)	223s	252s	−29s

Total across these five passing jobs: 773s (PR) vs 778s (main) — a 5-second / ~0.6% delta scattered in both directions, well within run-to-run noise.

Why the 200ms doesn't actually cost 200ms per test

The mutation-observer timer in Render.ts:90-95 only gates whenDOMUpdated() / renderFinished(). It fires once at the end of an update batch, not after every DOM write. And:

1. Tests that don't await renderFinished() (most Cypress assertions go through cy.get(...).should(...) retry loops) never wait on it at all.
2. Even tests that do await it get the timer re-armed on every new mutation — the 200ms delay is only paid once per quiet period, not per assertion.

So the 200ms is a "settle" guard, not a per-render wait. Closing this — the speedup hypothesis didn't hold.

Better targets if chasing test-speed wins

• The leading requestAnimationFrame in scheduleRenderTask (Render.ts:77) — headless Cypress browsers throttle rAF.
• The second rAF inside whenDOMUpdated (Render.ts:114).
• Per-test cy.wait(...) calls and explicit setTimeout polls in spec files.

[pskelin]

Closing — see comment above for the finding. The 200ms mutation-observer delay isn't a meaningful contributor to test duration.