[hoangchunghien]

Closes: #16701

stubtest failed on the stubsabot bump with 37 errors. 1.5.0 is a larger release than usual (two new calendars, parts of the day, new settings and several removals), so this updates the stubs to match:

• New public names: PartOfDay and PartsOfDay (re-exported from dateparser, defined in dateparser._parts_of_day), DateData.parts and DateData.part_of_day, and the STRICT_DATE_ORDER, TIMEZONE_ABBREVIATIONS, PARTS_OF_DAY and USE_GIVEN_LANGUAGE_ORDER settings.
• DateData.period can now be "quarter" or "part_of_day".
• search_dates() takes date_formats and add_period. The overloads cover the four tuple shapes that add_detected_language and add_period produce.
• New stubs for dateparser.calendars.minguo and dateparser.calendars.french_republican. The French Republican module imports convertdate, which stubtest does not install, so it joins hijri and jalali in the allowlist.
• pop_tz_offset_from_string() takes settings, with the same two overloads as the runtime.
• Removed upstream: Locale.to_parserinfo(), HijriDate.weekday(), _DateLocaleParser.parse() and _parser.parse(). Settings no longer caches instances, so the stub drops __new__.
• Marked @deprecated where 1.5.0 now emits a FutureWarning: Settings.get_key(), utils.registry(), convert_to_local_tz(), build_tz_offsets(), get_week_start(), get_week_end(), and calling generate_time_span() without base_date or settings.
• The _NonGregorianDate protocol uses read-only properties and drops weekday(), because the Minguo parser uses datetime.date and HijriDate no longer has weekday().

One new allowlist entry, PartOfDay.__new__: the runtime class is (str, Enum), so stubtest compares str.__new__ in the stub with Enum.__new__ at runtime.

tests/stubtest_third_party.py dateparser passes against 1.5.0; mypy_test.py, pyright_test.py and pre-commit pass for the package.

[github-actions]

According to mypy_primer, this change has no effect on the checked open source code. 🤖🎉