Add Undo/Redo support in the viewer

[emilk]

What

• Closes Undo (and redo) in viewer #3135
• Proceeded by Add ChunkStore::drop_time_range #7602
• Proceeded by ChunkStore: Protect specific time ranges during GC #7603

This PR implements Undo and Redo for any edit to the active blueprint.

undo.mp4

Implementation

This implements undo/redo by editing the "time cursor" for the blueprint timeline. Undo moves it backwards, redo forwards. When doing an action, all redo history is erased from the store with a new ChunkStore::drop_time_range function.

TODO

Closing a space-view and hitting undo fails to restore the space-view, and instead results in:

[WARN re_blueprint_tree::blueprint_tree] Bug: asked to show a UI for a space view that doesn't exist
[DEBUG egui_tiles::tiles] Failed to find tile #8930786982270384173 during simplify

The reason are these lines of code:

rerun/crates/viewer/re_viewport_blueprint/src/container.rs

Lines 338 to 343 in c599b56

	pub fn clear(&self, ctx: &ViewerContext<'_>) {
	ctx.command_sender.send_system(SystemCommand::DropEntity(
	ctx.store_context.blueprint.store_id().clone(),
	self.entity_path(),
	));
	}

rerun/crates/viewer/re_viewport_blueprint/src/space_view.rs

Lines 311 to 316 in c599b56

	pub fn clear(&self, ctx: &ViewerContext<'_>) {
	ctx.command_sender.send_system(SystemCommand::DropEntity(
	ctx.store_context.blueprint.store_id().clone(),
	self.entity_path(),
	));
	}

When clearing a container or blueprint, the entity is dropped from the store.
Just removing these lines fixes undo, but creates a problem of accumulating garbage, potentially wasting RAM as the blueprint grows (especially for users with thousands of entities).

Me and @jleibs came up with a very nice solution to this though:

• Remove the DropEntity calls above. Just keep the entities, but don't reference them
• Remove the GC flag that protects at least 1 copy of each entity
• Add a GC flag that protects everything newer than a certain timepoint (make sure we have e.g. 100 undo points)
  • ChunkStore: Protect specific time ranges during GC #7603
• Optional optimization: Add a EntityDb::clear_index to remove references to components/entities that have no data (have all been GCed)
• Optional optimization: Update this code to not walk the entity tree, but recursively collect entities from the root down

Checklist

• I have read and agree to Contributor Guide and the Code of Conduct
• I've included a screenshot or gif (if applicable)
• I have tested the web demo (if applicable):
  • Using examples from latest main build: rerun.io/viewer
  • Using full set of examples from nightly build: rerun.io/viewer
• The PR title and labels are set such as to maximize their usefulness for the next release's CHANGELOG
• If applicable, add a new check to the release checklist!
• If have noted any breaking changes to the log API in CHANGELOG.md and the migration guide

• PR Build Summary
• Recent benchmark results
• Wasm size tracking

To run all checks from main, comment on the PR with @rerun-bot full-check.

[emilk]

This is blocked by:

• Refactor viewport blueprint logic #8056