Parquet: read parquet metadata with page index in async and with size hints

[AdamGS]

Is your feature request related to a problem or challenge? Please describe what you are trying to do.

We have multiple cases where we want to just load parquet file metadata with the page index, we already have metadata and file sizes stored externally but there doesn't seem to be a way to read that data without using object_store.
fetch_parquet_metadata doesn't load the page index and but does use hints, and ArrowReaderMetadata doesn't use hints and forces us to jump through hoops and have an additional implementation of AsyncFileReader.

Describe the solution you'd like

I think that adding an argument to fetch_parquet_metadata is easier than stabilizing MetadataLoader, and will work with many other readers that can read byte ranges.

Describe alternatives you've considered

We currently just maintain a minor fork with MetadataLoader and MetadataFetch made public.

Will be glad to do any and all work required here (either adding an argument to fetch_parquet_metadata, helping stabilize MetadataLoader or anything else), but I would like to make sure I'm doing something that's actually valuable and not just spamming the maintainers here.

[tustvold]

I don't see a compelling reason not to make MetadataLoader public and would be happy to review a PR doing so, the API seems unlikely to change in the near future

[AdamGS]

Awesome! I'll hopefully have a PR ready today.

[tustvold]

label_issue.py automatically added labels {'parquet'} from #5137