add releasing docs, give action permission to write to the repo

.github/workflows/publish_release.yml

@@ -1,5 +1,8 @@
 name: Publish a release
 
+permissions:
+  contents: write
+
 on:
   workflow_dispatch:
     inputs:

README.md

@@ -36,9 +36,16 @@ Cloudera for assistance.  Additionally, further details on the AMP status can be
 
 ## Installation
 
+### Important
+
+#### The latest stable version of the AMP lives on the `release/1` branch. The `main` branch is the development branch and may contain unstable code.
+
 Follow the [standard instructions](https://docs.cloudera.com/machine-learning/cloud/applied-ml-prototypes/topics/ml-amp-add-catalog.html) for installing this AMP into your CML workspace.
 The "File Name" to use is `catalog-entry.yaml`.
 
+If you do not want to use the catalog-entry, then you should specify the release branch when installing the AMP directly:
+- `release/1` is the branch name to use for the latest stable release.
+
 For configuring the project, you will need to provide some environment variables.
 
 - `S3_RAG_BUCKET_PREFIX` - A prefix added to all S3 paths used by Rag Studio

RELEASING.md

@@ -0,0 +1,9 @@
+# Release process for the RAG Studio AMP
+
+* Get `main` into the state to be released
+* Publish a release from the `main` branch with the desired version to release
+* Verify the release in a CML workspace, deploying from `main`
+* Merge main into the `release/1` branch via a pull request, keeping the version file contents from main
+  * Let the PR build run, to verify no merge conflicts took place
+* If desired: publish a new release with the same version from the `release/1` branch to make sure the code matches the release bundles.
+