Wire Foundry IQ defaults into deployment parameters

[placerda]

Purpose

Make fresh GPT-RAG Zero Trust provisioning carry the Foundry IQ native Blob defaults into the Bicep/azd parameter layer and complete post-provision Foundry IQ knowledge source/base setup for standard extraction.

Depends on Azure/bicep-ptn-aiml-landing-zone#112. This PR currently pins the infra submodule to that feature-branch commit for validation. Before merge/release, cut the AI Landing Zone release and update .gitmodules, manifest.json, and the submodule pointer to the released tag.

Changes

• Add RETRIEVAL_BACKEND=foundry_iq and native FOUNDRY_IQ_* azd substitutions to main.parameters.json.
• Include non-empty Knowledge Base, connection, and Knowledge Source defaults so the landing-zone preflight can validate Foundry IQ deployments before ARM.
• Seed SEARCH_RAG_INDEX_NAME during postProvision and fix search.settings.j2 so dependent Foundry IQ defaults render before the key exists in App Configuration.
• Add FOUNDRY_IQ_AI_SERVICES_ENDPOINT support and render ingestionParameters.aiServices.uri plus matching chat model resourceUri when FOUNDRY_IQ_CONTENT_EXTRACTION_MODE=standard.
• Fail post-provision search setup instead of silently succeeding when templates cannot render.
• Add focused template tests for the fresh Foundry IQ rendering path.
• Pin the infra submodule to the AILZ branch commit that stamps SEARCH_RAG_INDEX_NAME and the AI Services endpoint for non-isolated App Configuration population.
• Update README and changelog operator notes.

Validation

• python -m unittest discover -s config/search/tests -v
• python -m compileall config/search/setup.py config/search/tests/test_foundry_iq_templates.py
• JSON parse check for main.parameters.json and infra/main.parameters.json
• az bicep build --file infra/main.bicep (passed with existing warnings)

Follow-up before full deploy

Release Azure/bicep-ptn-aiml-landing-zone#112 as the next AILZ tag, then update GPT-RAG to consume that tag instead of the feature-branch submodule commit.