xenopsd: Fix pytype warning on get_words: Name string is not defined

[bernhardkaindl]

For @stephenchengCloud:
FYI: @ashwin9390

In #5896 (review), changes were requested:

Before merging feature/py3 into master, certain warnings in files that weren't checked so far should be fixed.

This PR fixes two of the mentioned pytype warnings:

File "xen-api/ocaml/xenopsd/scripts/common.py", line 198, in get_words: Name 'string' is not defined [name-error]
File "xen-api/ocaml/xenopsd/scripts/common.py", line 201, in get_words: Name 'string' is not defined [name-error]

These uses of the not defined variable string are in an inner function of VIF.get_locking_mode().
This inner function (get_words) isn't used by the method, see:
https://github.com/xapi-project/xen-api/blob/master/ocaml/xenopsd/scripts/common.py#L196

The inner function was never used, it is completely bogus.

Because inner functions cannot be called from outside the containing method, it is 100% safe to remove it:

• Remove the unused inner function common.VIF.get_locking_mode().get_words() to fix pytype warnings and apply minor clean-ups after adding a test that covers all branches in the method, ensuring that the changes do not change functionality.
• Apply cosmetic formatting fixes to show the whole function in the review so it can be checked.

Two commits are to be added to this PR:`

• xenopsd: Add pytest file for common.VIF.get_locking_mode()
  • Passed before pushing the 2nd commit (see the blue checkmark for the commit)
• ocaml/xenopsd/scripts/common.py: Fix warnings: remove inner function, use isinstance
  • Test still passes: This ensures that there is no functional change to the changed method

[Vincent-lau]

this one is a subset of #5923 ?

[bernhardkaindl]

this one is a subset of #5923 ?

Yes, the first two commits of #5923 use the git commit hashes of this commit as base, because this fixes two of the pytype warnings that it needs to fix for enabling pytype checks on xenopsd's scripts.

Because the code changes in this PR are very minimal, I think we can merge this PR first and then #5923 becomes visually much smaller when it is rebased on the updated feature branch.