[DOCS] Document that read_blob() only supports CONTENT mode (DESCRIPTOR mode is not allowed)

**What needs to be done:**
Update the website docs (AI quick start and any other place `read_blob()` is referenced) to clearly state that `read_blob()` only supports the `CONTENT` mode. The `DESCRIPTOR` mode is not allowed and will fail at query time.

Concretely:
- Update the AI quick start guide (https://hudi.apache.org/docs/next/ai-quick-start-guide/) so all `read_blob()` examples use `CONTENT` mode.
- Fix the demo script that still uses `DESCRIPTOR` mode: [hudi-examples/hudi-examples-spark/src/test/python/vector_blob_demo/run_demos.sh](https://github.com/apache/hudi/blob/master/hudi-examples/hudi-examples-spark/src/test/python/vector_blob_demo/run_demos.sh).
- Add a short note in the BLOB / `read_blob()` reference docs calling out the supported modes explicitly so users don't hit the failure.

**Why this task is needed:**
Surfaced during 1.2.0-rc2 testing — running the existing vector/blob demos hit a failure because `DESCRIPTOR` mode is no longer allowed on `read_blob()` and the demos had to be changed to `CONTENT`. This is expected runtime behavior, but the docs and demos still reference the old usage, which will confuse users following the AI quick start.

Non-blocker for the 1.2.0 release; tracked as a docs follow-up.

Start with the AI quick start guide and the referenced hudi-examples/hudi-examples-spark/src/test/python/vector_blob_demo/run_demos.sh script, then locate the BLOB/read_blob() reference documentation. Update every mentioned example to use CONTENT mode and add the supported-mode note; done means no documented or demo usage still presents DESCRIPTOR as valid.

Written by the indexing model from the issue text.