Support basic semantic search with preset workflow template

[ohltyler]

Description

This PR updates several interfaces and improves upon some of the base building blocks to support end-to-end creation of a workflow template for basic semantic search, with heavy UI guardrails. Specifically:

• adds detailed hardcoded logic for toWorkspaceFlow() and toTemplateFlows() for a basic semantic search use case. toWorkspaceFlow() is used to load a static set of draggable nodes/edges on the UI. toTemplateFlows() is used to convert expected & validated field data (model ID, etc.) and populate a preset template configuration used for creating the actual workflow via create workflow API. Both of these will become more generic and flexible over time; this is to prove out an initial use case
• updates IComponentField schema to support more information for a particular field, including help text, links to external documentation, etc. UI components parse and render them accordingly
• onboards and integrates with the ML commons plugin, and the search model API, including integrating with redux store to persist fetched models on client-side. This is used for auto-populating available model IDs in the dropdown (see video below for example)
• minor update to ReactFlowProvider to move one layer up so ResizableWorkspace can use useReactFlow() hook. This was previously downscoped too far in the previous PR.
• simplifies the ReactFlow edge to use the preset ones provided from ReactFlow, and block deletion. This is part of the ongoing effort to downscope the amount of flexibility and user customization on the flow builder page for now
• integrates with ui_metadata field in the backend template to persist UI flow data
• updates template schema to support user_inputs and previous_node_inputs
• add optional provision field in createWorkflow() (currently not integrated into buttons yet)
• minor styling updates (preventing overflow scrollbars, vh tuning)

Screenshot of updated component inputs with help text and external links. Also includes the Existing/New tabs but scoped down to only allow New for now - this may change as scope for initial release is finalized.

Demo video showing end-to-end creation of a (mostly hardcoded) template configuration, provisioning the template, and executing semantic search with the newly-created ingest pipeline and knn index.

demo-e2e.webm

Issues Resolved

Makes progress on #73

Check List

• [x] Commits are signed per the DCO using --signoff

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license. For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 22.18%. Comparing base (3eb94a5) to head (4c109fe).

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #121 +/- ## ========================================== + Coverage 21.92% 22.18% +0.26% ========================================== Files 56 57 +1 Lines 812 870 +58 Branches 82 92 +10 ========================================== + Hits 178 193 +15 - Misses 630 673 +43 Partials 4 4 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[kamingleung]

What's the difference between Existing / New tabs? Can this be stateless?

[ohltyler]

What's the difference between Existing / New tabs? Can this be stateless?

Just what I have as a placeholder for now, with the idea that some components may allow using existing resources/data vs. creating new. Example being creating a new knn index or using an existing one, or creating a new model or using an existing one. This can easily be removed, in the current examples I have existing entirely disabled.