Column grouping layout hints are ignored on rollup tables

[mofojed]

A customer noticed that column groupings from layout hints are respected on regular tables, but ignored on rollup tables. (The columns can be grouped manually afterward and it works; the issue is just the layout hints being ignored.)

DH-17076

Only reproducible in Groovy:

import io.deephaven.engine.util.LayoutHintBuilder

source = emptyTable(100).update(
    "K1 = ii",
    "K2 = K1 % 10",
    "Y = randomDouble(0, 10)",
    "Z = randomDouble(0, 10)",
)

result_norollup = source.setLayoutHints(LayoutHintBuilder.get().columnGroup("MyCols", ["Y", "Z"]).build())

import io.deephaven.api.agg.spec.AggSpec
import io.deephaven.api.agg.Aggregation


result_rollup = source
    .rollup([Aggregation.of(AggSpec.avg(), "Y", "Z")], false, "K2")
    .setLayoutHints(LayoutHintBuilder.get().columnGroup("MyCols", ["Y", "Z"]).build())

For some reason, layout_hints is not an available method on RollupTable in Python: https://github.com/deephaven/deephaven-core/blob/dbf2b2db31ac3e982fc6c432b66d3cb688fda431/py/server/deephaven/table.py#L169

[mofojed]

Tried to translate the snippet to Core python:

from deephaven import empty_table, agg
source = empty_table(100).update(["K1 = ii", "K2=K1%10", "Y=randomDouble(0, 10)", "Z=randomDouble(0, 10)"])
result_norollup = source.layout_hints(column_groups=[{"name": "MyCols", "children": ["Y", "Z"]}])
result_rollup = source.rollup([agg.avg(["Y", "Z"])], "K2", False).layout_hints(column_groups=[{"name": "MyCols", "children": ["Y", "Z"]}])

However get an error because layout_hints is not a method on RollupTable. In Legacy, it appears HierarchicalTable extends QueryTable which extends BaseTable which has the layoutHints method, but in Core that's not the case.

So I don't think this can actually be reproduced in Core Python...

EDIT: It looks like HierarchicalTable extends GridAttributes, which does have the setLayoutHints method. It's just not accessible from Python, so this is reproducible from Groovy still. Not sure why not exposed in Python.