Triage which remaining 4.1 spans come back before 5.0

[heruan]

5.0 is not released, so no 4.1 span has actually been dropped. The consolidation decisions made during the rewrite were never reviewed as decisions, and this issue closes them out so that every 4.1 span has an explicit answer rather than an implicit omission.

Starting position, for argument:

Worth restoring

• Component creation (ComponentRenderer). A heavy per-row renderer is a common cause of a slow Grid, so this has real diagnostic value. Not as one span per cell, which would be trace spam. One span per render pass plus a timer tagged by component class. This needs a Flow SPI, so realistically after GA. Prior art: Restore renderer cost instrumentation lost in the 5.0 rewrite #362 already tracks restoring renderer cost instrumentation, so the decision here should land there rather than fork.
• A vaadin.view attribute on vaadin.rpc. 4.1 had it. Today route exists only on navigation spans, so an RPC cannot be attributed to a view. Cheap and fits in GA.

Proposed to keep consolidated

• Static resources, PWA resources and web component bootstrap. Low diagnostic value, and usually served by a CDN or reverse proxy in production. type=static already covers them.
• SessionRequestHandler.* and UnsupportedBrowserHandler.*. Negligible value.
• Per-handler RPC span names (Event:, Sync:, Invoke server method:, Attach existing element, Handle return channel, Handle navigation). vaadin.rpc with type, vaadin.rpc.event and vaadin.rpc.component carries more information than the 4.1 span names did.

Already covered

• Heartbeat, via type=heartbeat.

Tracked separately

Hilla endpoint spans and browser spans are scope decisions of their own, not part of this triage. Note that upgrading Hilla applications lose their endpoint spans with no replacement, which needs a release note line regardless of the outcome here.

Background: docs/spans-parity-plan.md.