chore: simplify component pop

[Rich-Harris]

couple of changes here:

• the only time an effect gets deferred is if it's a top-level $effect. when pop runs, the current active effect/reaction are the same as they were when the effect was declared, so the set_active_effect and set_active_reaction calls are no-ops. By extension, we don't need to stash the current active effect/reaction, and we don't need to wrap the effect creation in a try-catch
• since we don't need to stash the effect/reaction, component_context.e can just be an array of functions rather than an array of objects — less memory usage for components with effects
• we can identify a top-level $effect because the current active_effect is a BRANCH_EFFECT, and the current active_reaction is null. These conditions do not obtain in any other circumstances. Consequently, we can remove the m flag — it's unnecessary. This saves more memory

[changeset-bot]

🦋 Changeset detected

Latest commit: 97092d2

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
svelte	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[github-actions]

Playground

pnpm add https://pkg.pr.new/svelte@16331

[dummdidumm]

Since this changes a bite more we should add a change set