docs: add RAG / LLM failure-mode checklist tutorial

[onestardao]

Tracking issue

Related to #6930

Why are the changes needed?

RAG / LLM workloads on Flyte are becoming more common, but when a pipeline misbehaves it is still hard for users to reason about where the failure comes from (ingestion, embeddings, retriever, LLM, evaluation, etc.).

This PR adds a pointer to a concrete, MIT-licensed failure-mode checklist that:

• organizes 16 common failure modes across the main stages of a RAG / LLM pipeline
• is already referenced by external projects (e.g. ToolUniverse from Harvard MIMS Lab, QCRI LLM Lab’s Multimodal RAG Survey, and Rankify from the University of Innsbruck) as a debugging / evaluation aid
• can serve as a lightweight starting point for Flyte users who are building similar workflows and need a structured way to debug them

What changes were proposed in this pull request?

• Add a “RAG / LLM pipeline failure mode checklist” entry to the Tutorials section of the README.md.
• The link points to the WFGY ProblemMap page, which documents a 16-problem checklist for RAG / LLM pipelines (covering data ingestion and chunking, embeddings and vector stores, retrievers and ranking, LLM routing / tools, and evaluation / guardrails).

This is intentionally a minimal docs-only change.
As a follow-up, I am interested in contributing one or two Flyte 2.0 examples that use this checklist to structure RAG / LLM workflows, based on the suggestions in #6930.

How was this patch tested?

• Rendered the README using GitHub’s preview to verify:
  • the new tutorial entry renders consistently with existing links
  • the link target is correct and reachable
• No code or configuration changes were made, so no unit or integration tests were added.

Labels

• added: new documentation link / tutorial entry

Setup process

Not applicable – docs-only change.

Screenshots

Not applicable – README text-only change.

Check all the applicable boxes

• I updated the documentation accordingly.
• All new and existing tests passed.
• All commits are signed-off.

Related PRs

None.

Docs link

Not applicable – change is in the main README.md only.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 56.95%. Comparing base (b5f898c) to head (14e98fb).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #6931      +/-   ##
==========================================
- Coverage   56.96%   56.95%   -0.01%     
==========================================
  Files         929      929              
  Lines       58156    58156              
==========================================
- Hits        33130    33125       -5     
- Misses      21984    21989       +5     
  Partials     3042     3042              

Flag	Coverage Δ
unittests-datacatalog	53.51% <ø> (ø)
unittests-flyteadmin	53.10% <ø> (-0.04%)	⬇️
unittests-flytecopilot	43.06% <ø> (ø)
unittests-flytectl	64.02% <ø> (ø)
unittests-flyteidl	75.71% <ø> (ø)
unittests-flyteplugins	60.14% <ø> (ø)
unittests-flytepropeller	53.64% <ø> (ø)
unittests-flytestdlib	63.29% <ø> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[onestardao]

This PR is docs-only (just adding a link to the README). Codecov reports a -0.01% project coverage change coming from indirect changes in flyteadmin (workflow_execution_event_writer.go), but all modified lines in this PR are fully covered.

If needed, I’m happy to rebase so the checks can be re-run, but this looks like a coverage noise issue rather than a change introduced here.