ref(grouping): Use a dataclass for seer grouping request typing

[lobsterkatie]

This switches to using a dataclass for typing requests to seer for similar groups, primarily to make it easier to handle optional attributes.

In the long run, we might consider switching the typing for other associated types to dataclasses as well, for consistency, but to keep things simple for now I'm just switching this one since it's the only one with optional fields.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.85%. Comparing base (ff214bb) to head (59823b5).
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #69934      +/-   ##
==========================================
- Coverage   79.85%   79.85%   -0.01%     
==========================================
  Files        6490     6490              
  Lines      288620   288622       +2     
  Branches    49699    49700       +1     
==========================================
- Hits       230488   230484       -4     
- Misses      57734    57740       +6     
  Partials      398      398              

Files	Coverage Δ
...y/api/endpoints/group_similar_issues_embeddings.py	100.00% <100.00%> (ø)
src/sentry/seer/utils.py	100.00% <100.00%> (ø)

... and 4 files with indirect coverage changes

[lobsterkatie]

Closing this in favor of #70070.