Incremental fetching / pagination in the Results Viewer

[clang-engineer]

I'd like to suggest incremental fetching (or pagination / "Load More") for the Results Viewer.

Currently, when a hard fetch limit is enabled (for example, 1,000 rows), inspecting rows beyond that limit requires increasing the limit and executing the query again.

For exploratory use against large or remote databases, it would be useful if the Results Viewer could fetch additional rows on demand:

Execute query
    ↓
Fetch first 1,000 rows
    ↓
Results Viewer
    ↓
Load More
    ↓
Fetch next 1,000 rows

The exact implementation doesn't necessarily need to expose SQL LIMIT/OFFSET; ideally this could continue fetching from the underlying cursor where the adapter supports it.

I looked through the current implementation and noticed that HarlequinCursor currently exposes fetchall(), and query.fetch() drains the cursor into a ResultSet.

I also noticed that large-result streaming is already mentioned as future work requiring changes to the cursor interface. This seems closely related to #875 as well, since incremental fetching could help avoid materializing very large result sets up front.

Would incremental loading in the Results Viewer fit into that planned streaming work?

I'd be happy to work on this and submit a PR if it's something you'd be interested in supporting. If so, I'd appreciate guidance on the preferred direction for the cursor API (for example, fetchmany() / batch-oriented fetching) before starting the implementation.