gh-155175: Reject fractional seconds without a decimal mark in C fromisoformat

[SEPURI-SAI-KRISHNA]

parse_hh_mm_ss_ff() in Modules/_datetimemodule.c parsed any characters left
over after HHMMSS in the ISO 8601 basic format as a fractional seconds
component, without requiring the decimal mark that ISO 8601 mandates:

>>> time.fromisoformat('12345678')
datetime.time(12, 34, 56, 780000)          # now raises ValueError
>>> datetime.fromisoformat('2020-01-01T12345678')
datetime.datetime(2020, 1, 1, 12, 34, 56, 780000)   # now raises ValueError

The HH/MM/SS loop only breaks into the fraction-parsing code when it
sees . or ,. In the basic format it can also fall out of the loop normally
with input still unconsumed, the else if (!has_separator) { --p; } branch
un-consumes the character and the loop ends, after which the code below
parses whatever remains as a fraction, never checking that a decimal mark
introduced it.

This tracks whether a decimal mark was actually seen and rejects the string
otherwise, matching _pydatetime._parse_hh_mm_ss_ff(), which performs the
check explicitly:

if pos < len_str:
    if tstr[pos] not in '.,':
        raise ValueError("Invalid microsecond separator")

Because the same helper also parses the UTC offset, this fixes malformed
offsets being accepted with the trailing digits silently discarded:

>>> time.fromisoformat('12:34:56+00000000')
datetime.time(12, 34, 56, tzinfo=datetime.timezone.utc)   # now raises ValueError

Valid inputs are unaffected, '123456', '123456.78', '123456,78',
'123456.123456789', '12:34:56.123456', '1234', '12', '123456+0000',
'123456Z', '12:34:56+00:00:00.123456', '20200101T123456.789' and
'20200101T123456+0530' all parse exactly as before, and identically to the
pure-Python implementation.

This is the same class of C/pure-Python divergence as gh-152157 (empty fraction
before a timezone designator) and gh-152079.

Tests

Five cases added to each of the time and datetime bad_strs lists in
Lib/test/datetimetester.py, covering the fraction and the UTC offset paths.
They run under both the _Fast (C) and _Pure (pure Python) test classes.

test_datetime passes (1160 tests), and the full suite passes
(run=48,806, Result: SUCCESS).

• Issue: time/datetime.fromisoformat: C accelerator accepts a fractional seconds component with no decimal mark #155175

[python-cla-bot]

All commit authors signed the Contributor License Agreement.

[SEPURI-SAI-KRISHNA]

CIFuzz has no datetime target, Modules/_xxtestfuzz/fuzz_tests.txt lists 11 targets and none reach _datetimemodule.c. All three sanitizer variants failed identically, which points to an OSS-Fuzz infra issue rather than this change.

[aisk]

The comment and the NEWS entry are a bit too verbose IMO, but it's up to you. The failed CI should be unrelated, other PRs failed too. I restarted them.