feat: expose resetting run boundaries

[drake-nominal]

When creating a run, a user is required to enter in the start / end timestamps before:

• All datasets are added to the run
• All data is present in the datasets being added to the run (e.g. a dataset is mid ingest via multi-file or may otherwise be updated in the future)

As a result, customers can end up in a situation where the start / end bounds of a run aren't particularly accurate as data continues to be ingested into the platform, and having a simple way to just "reset" the bounds turns out to be powerful.

[alkasm]

All data is present in the datasets being added to the run (e.g. a dataset is mid ingest via multi-file or may otherwise be updated in the future)

IIUC this PR doesn't fix this issue? The datasets that are still mid-ingest are filtered out.

I'd like to minimize non-idempotent mutations as first-class functionality in the lib.

[drake-nominal]

IIUC this PR doesn't fix this issue? The datasets that are still mid-ingest are filtered out.

@alkasm in a single file world, sure, but this is a pretty rare edge case in the long run. Consider the case where the customer has 50000 files that compose one of the datasets instead-- now "mid ingest" can still mean that the dataset shows up as "ingested" in product. Or perhaps new files get added later after the run is created.