[ML] File upload: Allow import into existing index

[markwalkom]

Describe the feature: There's no way to import data into an existing index, but there are times that you may want to update existing documents (which might warrant another issue on how to handle) or add new data.

Describe a specific use case for the feature:
Let's say I get a monthly report with a few hundred lines, it makes more sense to have a yearly index and then just add each month to that, rather than an index per day/month.

[elasticmachine]

Pinging @elastic/ml-ui

[droberts195]

We deliberately limited import via the file data visualizer to a new index for three reasons:

1. We didn’t want to create a competitor to Beats/Ingest pipeline for repeated/production data import
2. It creates a whole host of decisions about what to do if the mappings clash for different files - if we demand they’re identical before importing that will lead to a request to coerce data that is “almost” the same
3. We didn’t want to risk messing up important existing indices due to typos in the index name - remember mappings can be added but never removed

Maybe more can be done around reusing an existing index but I think it needs broad discussion from a product management perspective as well as an isolated feature perspective.

[peteharverson]

See also: #49159