Perf regression: caching, single_query, updates

| Scenario | Environment | Date | Old RPS | New RPS | Change | Deviation | StDev | Dependencies |
| -------- | ----------- | ---- | ------- | ------- | ------ | --------- | ----- | ------------ |
| caching | Caching Platform 19- Trends Database cobalt-hosted-lin | 07/20/2026 09:07:50 | 339,424 | 163,643 | -51.79 % (-175,782) :thumbsdown: | -3.94 σ | 44,561.82 | Changes NameVersionDiff |
| single_query | Single Query Platform 19- Trends Database cobalt-hosted-lin | 07/20/2026 09:04:14 | 214,749 | 124,263 | -42.14 % (-90,486) :thumbsdown: | -3.29 σ | 27,498.41 | Changes NameVersionDiff |
| updates | Updates Platform 19- Trends Database cobalt-hosted-lin | 07/20/2026 09:06:45 | 31,917 | 22,218 | -30.39 % (-9,699) :thumbsdown: | -5.36 σ | 1,809.87 | Changes NameVersionDiff |

[PowerBI Dashboard](https://aka.ms/aspnet/benchmarks)

@sebastienros

Start with the linked PowerBI Dashboard and compare the caching, single_query, and updates benchmark runs dated 07/20/2026. Investigate the listed benchmark changes and dependencies to identify what changed between the old and new results. Done means locating the cause of the reported RPS regressions and confirming that the affected scenarios no longer show the regression.

Written by the indexing model from the issue text.