docs: release notes for curl-based s3

Author: lovesegfault

doc/manual/rl-next/s3-curl-implementation.md

@@ -0,0 +1,36 @@
+---
+synopsis: "Improved S3 binary cache support via HTTP"
+prs: [13752]
+issues: [13084, 12671, 11748, 12403, 5947]
+---
+
+S3 binary cache operations now happen via HTTP, leveraging `libcurl`'s native AWS
+SigV4 authentication instead of the AWS C++ SDK, providing significant
+improvements:
+
+- **Reduced memory usage**: Eliminates memory buffering issues that caused
+  segfaults with large files (>3.5GB)
+- **Fixed upload reliability**: Resolves AWS SDK chunking errors
+  (`InvalidChunkSizeError`) during multipart uploads
+- **Resolved OpenSSL conflicts**: No more S2N engine override issues in
+  sandboxed builds
+- **Lighter dependencies**: Uses lightweight `aws-crt-cpp` instead of full
+  `aws-cpp-sdk`, reducing build complexity
+
+The new implementation requires curl >= 7.75.0 and `aws-crt-cpp` for credential
+management.
+
+All existing S3 URL formats and parameters remain supported.
+
+## Breaking changes
+
+The legacy `S3BinaryCacheStore` implementation has been removed in favor of the
+new curl-based approach.
+
+**Migration**: No action required for most users. S3 URLs continue to work
+with the same syntax. Users directly using `S3BinaryCacheStore` class
+should migrate to standard HTTP binary cache stores with S3 endpoints.
+
+**Build requirement**: S3 support now requires curl >= 7.75.0 for AWS SigV4
+authentication. Build configuration will warn if `aws-crt-cpp` is available
+but S3 support is disabled due to an insufficient curl version.