[BUG] Fork log should error on incomplete manifest (#5771)

## Description of changes

_Summarize the changes made by this PR._

- Improvements & Bug fixes
- Throw error during `fork_log` if the first available offset in the
copied manifest is ahead of the compaction cursor
- New functionality
    - N/A

## Test plan

_How are these changes tested?_

- [ ] Tests pass locally with `pytest` for python, `yarn test` for js,
`cargo test` for rust

## Migration plan

_Are there any migrations, or any forwards/backwards compatibility
changes needed in order to make sure this change deploys reliably?_

## Observability plan

_What is the plan to instrument and monitor this change?_

## Documentation Changes

_Are all docstrings for user-facing APIs updated if required? Do we need
to make documentation changes in the_ [_docs
section](https://github.com/chroma-core/chroma/tree/main/docs/docs.trychroma.com)?_

Author: Sicheng-Pan

rust/log-service/src/lib.rs

@@ -1452,13 +1452,24 @@ impl LogServer {
             Arc::clone(&storage),
             target_prefix,
         );
+        let new_manifest = log_reader
+            .manifest()
+            .await
+            .map_err(|err| {
+                Status::new(
+                    err.code().into(),
+                    format!("Unable to read copied manifest: {}", err),
+                )
+            })?
+            .ok_or_else(|| Status::internal("Unable to find copied manifest"))?;
+        let first_copied_offset = new_manifest.oldest_timestamp();
+        if offset < first_copied_offset {
+            return Err(Status::internal(format!(
+                "Compaction cursor [{offset:?}] is behind start of manifest [{first_copied_offset:?}]",
+            )));
+        }
         // This is the next record to insert, so we'll have to adjust downwards.
-        let max_offset = log_reader.next_write_timestamp().await.map_err(|err| {
-            Status::new(
-                err.code().into(),
-                format!("Failed to read copied log: {}", err),
-            )
-        })?;
+        let max_offset = new_manifest.next_write_timestamp();
         if max_offset < offset {
             return Err(Status::new(
                 chroma_error::ErrorCodes::Internal.into(),