chore(hybrid-cloud): Resubmits Pydantic v2.7 upgrade

[GabeVillalobos]

Mulligan of PR #74770

This PR upgrades pydantic and updates the model configs we use for RPC methods to conform with the new version.

This PR does not address deprecation warnings, which will be handled in a series of follow-up commits to both Sentry and GetSentry.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.21%. Comparing base (2525223) to head (f3fa688).
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #75311      +/-   ##
==========================================
+ Coverage   78.18%   78.21%   +0.02%     
==========================================
  Files        6775     6775              
  Lines      302017   302019       +2     
  Branches    51963    51963              
==========================================
+ Hits       236134   236224      +90     
+ Misses      59513    59440      -73     
+ Partials     6370     6355      -15     

Files	Coverage Δ
src/sentry/autofix/utils.py	93.75% <100.00%> (ø)
src/sentry/hybridcloud/rpc/__init__.py	82.20% <100.00%> (-0.15%)	⬇️
src/sentry/hybridcloud/rpc/sig.py	90.78% <100.00%> (+0.24%)	⬆️
src/sentry/types/region.py	91.41% <100.00%> (+0.05%)	⬆️
src/sentry/users/services/user/serial.py	93.33% <ø> (ø)

... and 31 files with indirect coverage changes

[asottile-sentry]

Mulligan of PR #74770

any context on what's changed since last attempt? and why the last one failed?

[asottile-sentry]

the api diff also appears to be broken 🤔

[GabeVillalobos]

@asottile-sentry For some reason, a pod was OOMing in S4S after the upgrade. It's unclear why this was occurring as this particular pod shouldn't have been affected, but we're planning to try scaling the pod up a bit to see if this alleviates the issue.

As for the API diff errors, it looks like pydantic's generated JSON schema handles nullable fields differently in V2 than it did in V1 Pydantic docs on the changes

AppService.create_internal_integration_for_channel_request Schema

Old Schema:

"AppService__create_internal_integration_for_channel_request__ParameterModel": {
  "properties": {
    "organization_id": {"type": "integer", "title": "Organization Id"},
    "integration_name": {"type": "string", "title": "Integration Name"},
    "integration_scopes": {
      "items": {"type": "string"},
      "type": "array",
      "title": "Integration Scopes"
    },
    "integration_creator_id": {
      "type": "integer",
      "title": "Integration Creator Id"
    },
    "metadata": {"type": "object", "title": "Metadata"}
  },

New Schema

"AppService__create_internal_integration_for_channel_request__ParameterModel": {
  "properties": {
    "organization_id": {"type": "integer", "title": "Organization Id"},
    "integration_name": {"type": "string", "title": "Integration Name"},
    "integration_scopes": {
      "items": {"type": "string"},
      "type": "array",
      "title": "Integration Scopes"
    },
    "integration_creator_id": {
      "type": "integer",
      "title": "Integration Creator Id"
    },
    "metadata": {
      "anyOf": [{"type": "object"}, {"type": "null"}],
      "title": "Metadata"
    }
  },
  "type": "object",
  "required": [
    "organization_id",
    "integration_name",
    "integration_scopes",
    "integration_creator_id"
  ],
  "title": "AppService__create_internal_integration_for_channel_request__ParameterModel"
},

The new schema appears to be more accurate given the method definition:

def create_internal_integration_for_channel_request(
    self,
    *,
    organization_id: int,
    integration_name: str,
    integration_scopes: list[str],
    integration_creator_id: int,
    metadata: dict[str, Any] | None = None,
) -> RpcSentryAppInstallation:
    ...

[asottile-sentry]

cool just checking -- more memory makes sense to me given pydantic 2's move to a rust extension over cython so I wouldn't be surprised if a few workers need slight bumps (especially if they're running particularly hot)

[asottile-sentry]