feat: Capture node_data directly alongside feeback

[DafyddLlyr]

What does this PR do?

• Adds new node_data column to feedback table
• Writes to the node_data column when feedback is logged
• Rebuilds feedback_summary view to read from this column

Context

The current feedback_summary view in pretty slow running, likely due to the filtered join per-row to the published_flows table. This solution stores the node data directly alongside the feedback, thus mitigating this issue (hopefully!).

Local testing is good - significantly faster to query the view now.

Next steps...

Once the following changes are live on production, I'm proposing to run the following SQL to populate the new column. I'm opting to do this manually instead of as a Hasura migration as it will be slightly easier to manage and monitor, and it's pretty low-risk (we're just populating a new column).

UPDATE feedback
SET node_data = (
    SELECT 
        (published_flows.data -> feedback.node_id) -> 'data'
    FROM 
        published_flows
    WHERE 
        published_flows.flow_id = feedback.flow_id 
        AND published_flows.created_at < feedback.created_at
    ORDER BY 
        published_flows.created_at DESC
    LIMIT 1
);

I've tested this locally with a copy of feedback data, however the sync script just copies a single published flow so it only populates a few records.

This matches the current join on the flow, as seen here -

planx-new/hasura.planx.uk/migrations/1715784133713_run_sql_migration/up.sql

Lines 37 to 49 in 4eeea8f

	LEFT JOIN LATERAL
	(
	SELECT
	(published_flows.data -> fb.node_id) -> 'data' AS data
	FROM
	published_flows
	WHERE
	published_flows.flow_id = fb.flow_id
	AND published_flows.created_at < fb.created_at
	ORDER BY
	published_flows.created_at DESC
	LIMIT 1
	) AS published_flow_node ON true;

[github-actions]

🤖 Hasura Change Summary compared a subset of table metadata including permissions:

Updated Tables (1)

• public.feedback permissions:

  	insert	select	update	delete
  public	➕/➖

  1 added column permissions

  	insert	select	update
  public	➕ node_data

[github-actions]

Removed vultr server and associated DNS entries

[jessicamcinchak]

Code looks good & plan sounds solid! Agree since it's a new column, it's an easy to manage manual migration for existing records (with comfort we can remove/retry if any surprises) 👍

[jessicamcinchak]

💯

[DafyddLlyr]

Update: SQL executed on production, getting expected results and feedback_summary view now significantly faster and loading in Metabase without issue 🎉