Support writing Parquet encoding stats

[alexjo2144]

Encoding stats are used by the reader to check if the dictionary pages can be used for predicate pushdown.

Fixes #9554

TODO: Still need to add a test, but I did manually test that there's dictionary based pushdown happening now that didn't happen before.

[findepi]

lgtm. is it testable?

[findepi]

how small can the smallest page be? can we overflow here?

[alexjo2144]

The field this in PageEncodingStats this gets used for is also an int, so it should be fine.

[findepi]

Should V2 here change in #9497?
if so, let's add a comment it in that PR... (also, would be yet another reason to simplify and not try to have a toggle there)

cc @losipiuk

[findepi]

Issue: #9554

nit: in the PR description change this to "Fixes #9554" to have github link the two.

[alexjo2144]

For testing, I can add a test with something like:
CREATE TABLE test_table (n) WITH (format = 'parquet') AS VALUES 1, 1, 2, 2, 4, 4, 5, 5

And assert that SELECT count(*) FROM test_table WHERE n = 3 reads no rows, if that's worthwhile. I'm not sure I can write a lower level test easily though.

[findepi]

And assert that SELECT count(*) FROM test_table WHERE n = 3 reads no rows, if that's worthwhile. I'm not sure I can write a lower level test easily though.

better such test than nothing, imo

[alexjo2144]

And assert that SELECT count(*) FROM test_table WHERE n = 3 reads no rows, if that's worthwhile. I'm not sure I can write a lower level test easily though.

better such test than nothing, imo

Added

[findepi]

test (:trino-accumulo) failure seems unrelated, but we likely don't have an issue for that

Error:  io.trino.plugin.accumulo.TestAccumuloConnectorTest.init  Time elapsed: 13.082 s  <<< FAILURE!
java.lang.RuntimeException: org.apache.accumulo.core.client.impl.AccumuloServerException: Error on server localhost:9997

[findepi]

CI #9617