[fitsbits] INHERIT and Hierarchical Grouping

[Rob Seaman]

> Perhaps this EXTTYPE convention should be written up and submitted to
> the Registry of conventions.  Even though this convention only  
> defines a
> single new keyword (EXTTYPE), it is a rather fundamental concept  
> that is
> relevant to many FITS files, and it also has implications on how the
> reserved EXTNAME keyword should be used.

Conventions should be documented, no matter how tersely.

However, shouldn't this rather be pursued as an addition to the  
standard itself?  One would hope that one effect of the new registry  
would be to highlight proposals upon which consensus is already  
high.  The checksum proposal could similarly move forward within the  
standards process - if nobody has suggested an alternative in the  
last decade, they aren't likely to do so now.

Rob