ref(explorer): Set reasoning_effort to medium for explorer chat

[JoshFerge]

Summary

• Sets reasoning_effort="medium" on the interactive explorer chat SeerExplorerClient, up from the seer-side default of "low"
• Paired with getsentry/seer#5775 which adds reasoning_effort as a dimension in satisfaction metrics so we can measure the impact

Note on bugbot comment: reasoning_effort is only sent on start_run, not continue_run. This is fine because seer persists reasoning_effort in the ExplorerState at start_run time and reads it from the stored state on subsequent requests — continue_run does not read reasoning_effort from the request payload.

Test plan

• Monitor satisfaction metrics after deploy, faceting by reasoning_effort to compare medium vs low

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit e698275. Configure here.}