`pystac.read_file` returns temporal extents with `tzinfo=tzlocal()` depending on the environment.

[TomAugspurger]

Depending on the environment (I think the value of TZ, maybe other things), the timezones on pystac Collections can be tzlocal rather than utc.

>>> pystac.read_file("https://planetarycomputer.microsoft.com/api/stac/v1/collections/sentinel-2-l2a").extent.temporal.intervals[0][0].tzinfo
tzlocal()

Here's a demonstration of the environment TZ influencing the output:

❯ TZ=UTC python -c 'import pystac; print(pystac.read_file("https://planetarycomputer.microsoft.com/api/stac/v1/collections/sentinel-2-l2a").extent.temporal.intervals[0][0].tzinfo)'
tzlocal()

❯ TZ="US/Central" python -c 'import pystac; print(pystac.read_file("https://planetarycomputer.microsoft.com/api/stac/v1/collections/sentinel-2-l2a").extent.temporal.intervals[0][0].tzinfo)'
tzutc()

Since the STAC spec says that all datetimes should be UTC, I think that the datetimes should be tz-aware. If python-dateutil happens to return a tzlocal timezone, I think pystac can assume that local = UTC and localize it to UTC.

Upstream issue in python-dateutil: dateutil/dateutil#349

xref #760. I'm not sure if this would fix this behavior or not (cc. @philvarner)