Fix/gcp etag issue

[zoryamba-elastio]

Part of: https://github.com/elastio/elastio/issues/11400

This PR contains rebase of existing changes to last release tag of mountpoint-s3 v1.23.0.

Also it updates aws-c-s3 submodule to fix the GCP issue.

See actual commit meassage describing the fix.

[zoryamba-elastio]

So, about the fix itself:

We're removing goto statement, which skips 4 blocks, responsble for:

1. buffer size optimization if (auto_ranged_get->force_dynamic_part_size)
2. response checksum validation if (meta_request->checksum_config.validate_response_checksum)
3. connection limits for s3express buxkets if (meta_request->is_express...)  meta_request->max_active_connections_override = ...
4. handling customer-specified headers_callback if (!empty_file_error && meta_request->headers_callback != NULL)

So, nothing critical here...

At the moment we're getting error because this statement is skipped: found_object_size = true;
Then we don't hit this statement: auto_ranged_get->synced_data.object_range_known = true;
and infinitely hit this statement goto has_work_remaining;
so the meta request is always pushed back to list aws_linked_list_push_back(...)

but the object size is definitely found, otherwise we should goto even earlier here: if (s_discover_object_range_and_size(...))

Thus this specific goto statement looks more like a bug, which not was discovered by maintainers yet, because they only test on AWS API