Added strict flag for clean_prefix_map()

[hrshdhgd]

Fixes #351

As of the current state, if all prefixes in a dataframe are not accounted for in the curie_map, no output is generated and the process errors out.

For now setting the default to be True but I'm not sure if it should be True/False. Thoughts @matentzn ?

[hrshdhgd]

@matentzn , this is ready for review based on our discussion here

[matentzn]

This PR is a no-brainer and can be merged, but it wont solve the OAK problem -the question is who is responsible for filling in missing prefixes?

[hrshdhgd]

We could do it here using your solution using the missing IRI but does that mean when the missing IRI is added to the curie_map, all prefixes that aren't accounted for get an additional prefix?

For e.g. let's say ABCD:1234 is unaccounted for. Now curie_map will look like

curie_map:
     BIRNLEX:  http://w3id.org/oak/unknown_prefixes/birnlex

and all mentions of ABCD:XXXX in the dataframe gets converted into BIRNLEX: ABCD:XXXX ?

Assuming it won't be oak in the IRI but something else.

[matentzn]

I was thinking this:

For e.g. let's say ABCD:1234 is unaccounted for. Currently curie_map looks like

curie_map:
     MONDO:  http://purl.obolibrary.org/obo/MONDO_

Now what you do is this inject the ABC prefix:

curie_map:
     MONDO:  http://purl.obolibrary.org/obo/MONDO_
     ABCD:  http://w3id.org/sssom/unknown_prefix/abcd/

Thats it?

[hrshdhgd]

Aaahhh .... much elegant. I likes!

[hrshdhgd]

But does that mean the strict flag would be useless at that point?

[matentzn]

Definetely not. this should only be done in strict=false mode.