load_tif_python ignores a provided array and crashes when called without one

Nobody has claimed this yet.

Hi, I've been reading through the Python fallback readers in
leapctype.py and noticed something in load_tif_python that looks
like a typo:

if x is not None:
x = np.zeros((numRows, numCols), dtype=np.float32)
elif x.shape[0] != numRows or x.shape[1] != numCols:
return None

If you pass an array in `x` it gets silently overwritten with a
fresh zero array, and if you don't pass one it crashes on
`x.shape` since x is None. Should be `if x is None:` so the shape
check applies to a passed-in array. The other readers in the file
(load_data for example) use the is None pattern, so this one just
looks inverted. Happy to open a PR if that'd help.

Start in leapctype.py at the load_tif_python entry point and compare its handling of x with the load_data reader mentioned in the issue. Verify that both calls with a supplied array and calls without one complete correctly, while preserving the shape check for supplied arrays.

Written by the indexing model from the issue text.