Table reading/writing to/from fits does not support writing 1d arrays with a single element #10765

Description

Astropy Table writing and reading to/from fits looses the information about dimensionality,
if the cell data written is a 1d array with a single entry.
Then, on reading back the data, it will be read as a number, so ndim=0, instead of ndim=1.

Expected behavior

Preserve this informatino

Actual behavior

Wrong dimensionality

Steps to Reproduce

❯ python
Python 3.8.5 (default, Sep  5 2020, 10:50:12) 
[GCC 10.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from astropy.table import Table
>>> t = Table({'x': [[1], [2]]})
>>> t.write('/tmp/test.fits', overwrite=True)
>>> t['x'].shape
(2, 1)
>>> Table.read('/tmp/test.fits')['x'].shape
(2,)

The error seems to be in both reading and writing.

The writing part fails to add the TDIM keyword and the reading part has a regex that does not match a tuple with just one entry.

System Details

Linux-5.8.10-1-MANJARO-x86_64-with-glibc2.2.5
Python 3.8.5 (default, Sep  5 2020, 10:50:12) 
[GCC 10.2.0]
Numpy 1.19.2
astropy 4.0.1.post1
Scipy 1.5.2
Matplotlib 3.3.2