pygmt.blockm*: Add 'output_type' parameter for output in pandas/numpy/file formats

[seisman]

Changes in this PR:

1. Use virtualfile instead of tempfile for output
2. Add the output_type parameter for pandas/numpy/file outputs [This is a new parameter for this wrapper, so the PR is marked as "enhancenment"
3. Add type hints for return value
4. Add type hints for output_type/outfile parameters

Preview:

• https://pygmt-dev--3103.org.readthedocs.build/en/3103/api/generated/pygmt.blockmean.html#pygmt.blockmean
• https://pygmt-dev--3103.org.readthedocs.build/en/3103/api/generated/pygmt.blockmedian.html#pygmt.blockmedian
• https://pygmt-dev--3103.org.readthedocs.build/en/3103/api/generated/pygmt.blockmode.html#pygmt.blockmode

Benchmarks:

>>> import pygmt
>>> data = pygmt.datasets.load_sample_data(name="bathymetry")
>>> for inc in ["01m", "30s", "15s", "5s"]:
...     %timeit data_bmean = pygmt.blockmean(data=data, region=[245, 255, 20, 30], spacing=inc)
...

Main branch:

111 ms ± 152 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)
191 ms ± 1.94 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)
297 ms ± 445 µs per loop (mean ± std. dev. of 7 runs, 1 loop each)
804 ms ± 12.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

This branch:

20.9 ms ± 143 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)
52.3 ms ± 174 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)
141 ms ± 2.44 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)
635 ms ± 12.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)