UpdateLog calls from watcher intermittently fail with canceled context; no UpdateLog call error from watcher are re-reconciled

The optional storage of underlying Pod logs for PipelineRuns or TaskRuns would intermittently fail because the golang Contex used for remote procedure calls t was not properly managed in the Tekton Results Watcher. In both the Watcher and Tekton Results API Server the user would see messages about the 'UpdateLog' action failing with 'canceled context error' The fix for this introduced proper context coordination in the Watcher's update logs feature. This includes adding a default timeout for update log operations of 30 seconds. That default can be modified by setting in seconds the 'update_log_timeout' argument on the Watcher's deployment. For example, "60s" would change to 60 seconds.