Fix disk index search deadlock when using 1 thread

[sunpenghao]

• Does this PR have a descriptive title that could go in our release notes?
• Does this PR add any new dependencies?
• Does this PR modify any existing APIs?
  • Is the change to the API backwards compatible?
• Should this result in any changes to our documentation, either updating existing docs or adding new ones?

Reference Issues/PRs

Fixes #462

What does this implement/fix? Briefly explain your changes.

Several steps during disk index search initialization read graph nodes into memory by "borrowing" search thread data to issue async IOs, and #417 wrapped this process into read_nodes. However, the original call sites are not cleaned so the thread data are borrowed repetitively. With -T 1, there is only one thread data item, and initialization will therefore deadlock on waiting for available thread data.

Fix this by removing unneeded thread data borrowing.

Any other comments?

N/A