Performance regression since resource as components

## Bevy version and features

main since https://github.com/bevyengine/bevy/pull/20934

## What you did

```
cargo run --release --example bevymark -- --waves 60 --per-wave 500 --benchmark --mode mesh2d
```

## What went wrong

Confirmed also on windows:

```
7950x, 4070ti, windows
cargo run --release --example bevymark -- --waves 60 --per-wave 500 --benchmark --mode mesh2d
13.7ms Before Resource as Components
15.3ms After Resource as Components
```

Start by reviewing the changes in PR #20934, then run the reported release-mode bevymark command with mesh2d, 60 waves, and 500 entities per wave. Compare the benchmark with the resource-as-components change on the current main branch and trace the affected performance path. Done means identifying and fixing the regression while preserving the benchmark behavior.

Written by the indexing model from the issue text.