DRAGEN: use sample name cleaning patterns instead of hardcoded ones

[vladsavelyev]

Add fn_clean_exts for DRAGEN sample name cleaning, instead of hardcoding patterns into the code.

Fixes #1865

[ewels]

Suggestion is to have a top level config attribute called dragen_config with the search patterns under there, rather than changing the search pattern config structure.

So something like:

dragen_config:
  filename_suffixes:
    - _tumor

[vladsavelyev]

Actually, there is no need to have _tumor and _normal as the cleaning patterns at all, since s_name = s_name.split(ext["pattern"], 1)[0] strips a suffix even if it's in the middle of a sample name (that feels even more dangerous, though!).

So removing merging the list of DRAGEN cleaning patterns with the main least, and removing _tumor and _normal from it.

[vladsavelyev]

@ewels, good to take another look now.

[ewels]

Awesome, thanks 👌🏻 It was a good discussion earlier, I'm still a little on the fence! Let's keep it in the back of our minds when we do some of the heavier refactoring in v2. Happy to reconsider the best approach.

I think there are a couple which are maybe a bit risky, in that I could imagine them legitimately coming up in some sample names (.time, .gc, .vc) - would it work to shift them to the other config where they're only removed from the end of sample names and not the middle onwards? I've done that to compromise with other similarly risky ones in the past.