fix(clickhouse): combine physical_properties into one SETTINGS clause

[mokashang]

Description

Fixes #5803.

A ClickHouse model that declared more than one physical_properties entry produced SQL with repeated SETTINGS keywords, e.g.

SETTINGS min_age_to_force_merge_seconds = 3600 SETTINGS min_age_to_force_merge_on_partition_only = 1

ClickHouse rejects this with a syntax error. The valid form is a single comma-separated clause:

SETTINGS min_age_to_force_merge_seconds = 3600, min_age_to_force_merge_on_partition_only = 1

Root cause

In ClickhouseEngineAdapter._build_table_properties_exp (and the analogous _build_view_properties_exp), every leftover table_properties_copy / view_properties_copy entry was wrapped in its own exp.SettingsProperty and added to the property list. sqlglot renders each SettingsProperty as a separate SETTINGS … keyword, so N settings became N independent clauses.

Fix

Refactor _build_settings_property to accept the full settings mapping and emit a single SettingsProperty whose expressions list holds one exp.EQ(...) per key=value pair. Both the table and view paths now call it once with the entire dict and append the resulting property, producing the expected SETTINGS k1 = v1, k2 = v2, … output.

The behavior for a single setting is unchanged — a one-entry SettingsProperty still renders as SETTINGS k = v.

Test Plan

• Updated the existing test_model_properties assertion that pinned the broken SETTINGS prop1 = 1 SETTINGS prop2 = '2' output; it now expects the combined form.
• Added a three-entry regression test inside test_model_properties that mirrors the model from the issue report (min_age_to_force_merge_seconds, min_age_to_force_merge_on_partition_only, index_granularity) to guard against the failure mode beyond N=2.
• Added a new test_view_properties_combine_settings that exercises _build_view_properties_exp directly, since the same bug existed there and had no unit-test coverage.
• Verified the rendered SQL with sqlglot matches ClickHouse's accepted syntax.
• pytest tests/core/engine_adapter/ — 517 passed, 1 skipped, no regressions.
• ruff check and mypy clean on the modified files.

Checklist

• I have run make style and fixed any issues
• I have added tests for my changes (if applicable)
• All existing tests pass (make fast-test)
• My commits are signed off (git commit -s) per the DCO