refactor: use datetime/timedelta for exponential backoff elapsed time

[TrueFurina]

Summary

Per #1336 (and the maintainer's suggestion in the thread): swaps the raw time.time() / monotonic-elapsed arithmetic in tableauserverclient/exponential_backoff.py for datetime/timedelta.

Changes

• tableauserverclient/exponential_backoff.py:
  • self.start_time = time.time() → self.start_time = datetime.now()
  • elapsed = time.time() - self.start_time → elapsed = (datetime.now() - self.start_time).total_seconds()
  • time.sleep(...) retained (sleep still needs seconds)

Behavior is unchanged: elapsed is still a float number of seconds, timeout comparison and error message are identical.

Verification

• Syntax check passed
• Timer still sleeps with backoff and raises TimeoutError when the deadline passes (verified locally with a 0.01s timeout)

[salesforce-cla]

Thanks for the contribution! Before we can merge this, we need @TrueFurina to sign the Salesforce Inc. Contributor License Agreement.

[jacalata]

Hi @TrueFurina — apologies, my earlier "PRs welcome" note on #1336 was too optimistic. Looking closely: (datetime.now() - self.start_time).total_seconds() works the same way as time.time() - self.start_time; nothing downstream uses timedelta types; both are equally subject to wall-clock jumps. Closing this and updating the issue. Really sorry for the wasted effort.