Fix cascading update guard diagnostics

[ryanflorence]

The scheduler's cascading update guard used one per-turn counter for all component updates. That meant a large batch of legitimate clientEntry hydration/update work could look like an infinite loop and stop later components from becoming interactive.

This keeps the performance signal while making the hard guard more precise. The runtime now warns once per event-loop turn when aggregate cascading component updates cross the threshold, but only dispatches the infinite-loop error when the same component instance repeatedly updates itself.

Closes #11222.

• Adds component-name/count diagnostics to the aggregate warning and hard error
• Tracks the hard error limit per component instance instead of globally
• Preserves the existing protection for handle.update() loops during render
• Adds a regression for 60 component instances updating in the same event-loop turn

[github-actions]

Preview Build Available

A preview build has been created for this PR. You can install it using:

pnpm install "remix-run/remix#preview/pr-11577&path:packages/remix"

This preview build will be updated automatically as you push new commits.

[ryanflorence]

@sbaleno mind trying this patch out for #11222?

[sbaleno]

@ryanflorence All good here now - thank-you