Fill with -1 instead of nan when dtype is integer

[fanurs]

Hi! While learning about the prototype, I encountered an error when running the Data Structures Colab notebook:

The error comes from the following line:

cngi_prototype/cngi/_utils/_table_conversion2.py

Line 365 in 920580a

data = np.stack([np.pad(rr[col], [(0,ss) for ss in mshape - np.array(rr[col]).shape], 'constant', constant_values=np.nan) for rr in tr])

Specifically, because assoc_spw_id has a dtype of integer, it is impossible to fill it with np.nan.

I am not very familiar with the context of these data, but a quick fix would be to fill it with, say, -1 (in this PR), or to simply convert the dtype into float.