Reviewed the diff and did a smoke test opening a working data directory and
stepping through the GUI — everything behaves as expected.

The changes are carefully largely mechanical (== None → is None,
type(x) == Y → type(x) is Y, == True → truthy, unused import removal,
DataFrame/Series → pd.DataFrame/pd.Series) and I verified each of
the removed imports is no longer referenced in the file. Two changes are
arguably latent bug fixes: scipy.mean → mean (scipy isn't actually
imported at the top level and scipy.mean was removed in SciPy 2.0), and
math.isnan → np.isnan (math isn't imported either).

The only behavioral tightening is the bare except: → except FileNotFoundError:
changes in get_data_info and update_pmag_tables. My smoke test used a
complete working directory so it didn't exercise those paths — worth noting
that read_magic_file and pmag.magic_read both handle missing files
internally and return [] rather than raising, so the except FileNotFoundError:
branches will effectively never fire regardless. If anything, it's better
to let unexpected exceptions propagate than silently swallow them, so the
tightening is a net positive.

A couple of small follow-up opportunities spotted while reading:

• Lines 3636 and 3652 have type(tmin) is str or type(tmin) is str and tmin != '' —
  the two conditions are identical, so the whole expression reduces to
  type(tmin) is str.
• The file has both import numpy as np and from numpy import vstack, sqrt, arange, array, mean, nan, which invites inconsistent np.array vs
  array usage. Worth consolidating to one style in a separate PR.
• A leftover #DEBUG print in add_fit (line 2586) that probably makes sense to look into and clean-up.