Perf regression: antiforgery-noop, json_middleware

| Scenario | Environment | Date | Old RPS | New RPS | Change | Deviation | StDev | Dependencies |
| -------- | ----------- | ---- | ------- | ------- | ------ | --------- | ----- | ------------ |
| antiforgery-noop | Antiforgery No Operation 13- Trends Cobalt Hosted Lin Server | 10/05/2025 19:43:18 | 554,240 | 280,800 | -49.34 % (-273,440) :thumbsdown: | -2.83 σ | 96,644.27 | Changes NameVersionDiff Microsoft.AspNetCore.Antiforgery.dll `10.0.0-rtm.25479.102` -> `10.0.0-rtm.25479.115` https://github.com/dotnet/dotnet/compare/660fe377a3020394d07f93ac3ccd94fa608ffe12...e72b5bbe719d747036ce9c36582a205df9f1c361 Microsoft.AspNetCore.App `10.0.0-rtm.25479.102` -> `10.0.0-rtm.25479.115` https://github.com/dotnet/aspnetcore/compare/660fe377a3020394d07f93ac3ccd94fa608ffe12...e72b5bbe719d747036ce9c36582a205df9f1c361 Microsoft.NETCore.App `10.0.0-rtm.25479.102` -> `10.0.0-rtm.25479.115` https://github.com/dotnet/runtime/compare/660fe377a3020394d07f93ac3ccd94fa608ffe12...e72b5bbe719d747036ce9c36582a205df9f1c361 |
| json_middleware | Json 10k 13- Trends Cobalt Hosted Lin Server | 10/05/2025 19:42:22 | 266,187 | 163,028 | -38.75 % (-103,159) :thumbsdown: | -2.43 σ | 42,461.13 | Changes NameVersionDiff Microsoft.AspNetCore.Antiforgery.dll `10.0.0-rtm.25479.102` -> `10.0.0-rtm.25479.115` https://github.com/dotnet/dotnet/compare/660fe377a3020394d07f93ac3ccd94fa608ffe12...e72b5bbe719d747036ce9c36582a205df9f1c361 Microsoft.AspNetCore.App `10.0.0-rtm.25479.102` -> `10.0.0-rtm.25479.115` https://github.com/dotnet/aspnetcore/compare/660fe377a3020394d07f93ac3ccd94fa608ffe12...e72b5bbe719d747036ce9c36582a205df9f1c361 Microsoft.NETCore.App `10.0.0-rtm.25479.102` -> `10.0.0-rtm.25479.115` https://github.com/dotnet/runtime/compare/660fe377a3020394d07f93ac3ccd94fa608ffe12...e72b5bbe719d747036ce9c36582a205df9f1c361 |

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

@sebastienros

Start by reviewing the antiforgery-noop and json_middleware benchmark scenarios and comparing the linked dotnet, aspnetcore, and runtime commits. Reproduce the reported RPS changes using the benchmark setup referenced by the PowerBI Dashboard, then identify the regression and confirm that both scenarios return to expected performance.

Written by the indexing model from the issue text.