Auto preagg widgets and inline SQL metrics

[nicosuave]

Summary:

• auto-materialize widget preaggregations from query intents
• allow inline SQL aggregate metrics without explicit agg
• update widget demo notebook with auto-preagg example

Tests:

• uv run ruff format --check . --exclude docs/_extensions --exclude sidemantic-duckdb/extension-ci-tools --exclude sidemantic-duckdb/scripts --exclude sidemantic-duckdb/duckdb --exclude sidemantic/adapters/malloy_grammar

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 8d3fbdf256

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[chatgpt-codex-connector]

Guard inline-agg SQL that uses unqualified columns

Allowing measure.agg is None when _sql_has_aggregation() is true means metrics like sql: "SUM(price) / SUM(cost)" now pass validation, but SQLGenerator._extract_metric_filter_columns only adds columns to the model CTE when they are table-qualified (it checks col.table), so unqualified price/cost references are skipped and the generated query will hit missing-column errors when it runs. This only happens for inline-aggregate SQL that doesn’t use {model}./qualified column names; consider enforcing qualification here or broadening column extraction to include unqualified column references for these metrics.

Useful? React with 👍 / 👎.