Fix unstable output of periodic sequence due to precision error.

[shunping]

A fix following #35300.

See the explanation below (copied from the inline comments):

In ImpulseSeqGenRestrictionProvider, the total number of counts (i.e. total_outputs) is computed by ceil((end - start) / interval) (

beam/sdks/python/apache_beam/transforms/periodicsequence.py

Line 45 in facdc4e

total_outputs = math.ceil((end - start) / interval)

), where end is start + duration (

beam/sdks/python/apache_beam/ml/ts/util.py

Line 172 in facdc4e

stop = start + min(self._duration, self._max_duration)

).

Due to precision error of arithmetic operations, even if duration is set to len(self._data) * interval, (end - start) / interval could be a little bit smaller or bigger than len(self._data).

In case of being bigger, total_outputs would be len(self._data) + 1, as the ceil() operation is used, which is NOT what we want.

Assuming that the precision error is no bigger than 1%, by subtracting a small amount, we ensure that the result after ceil() is stable even if a certain degree of precision error is present.

[shunping]

@claudevdm, could you please review this one too since you did the previous one and have some context? Thanks a lot!

[github-actions]

Assigning reviewers:

R: @jrmccluskey for label python.

Note: If you would like to opt out of this review, comment assign to next reviewer.

Available commands:

• stop reviewer notifications - opt out of the automated review tooling
• remind me after tests pass - tag the comment author after tests pass
• waiting on author - shift the attention set back to the author (any comment or push by the author will return the attention set to the reviewers)

The PR bot will only process comments in the main thread (not review comments).

[shunping]

stop reviewer notifications

[shunping]

remind me after tests pass

[github-actions]

Stopping reviewer notifications for this pull request: requested by reviewer. If you'd like to restart, comment assign set of reviewers

[shunping]

Thanks @claudevdm!