Pn/another interpretation refactor

[psnairne]

This PR is broken, because I have not implemented the methods on GIBuilder yet.

What does this PR do?

1. Makes collection of interpretations more similar to the other collector functions.
2. Creates a GenomicInterpretationBuilder struct which will handle the building of GenomicInterpretations. Previously this duty was part of PIVOTAL, but the problem with that is that we need to build GenomicInterpretations in a way that is quite specific to our project. So I am going to remove it from PIVOTAL, and transfer it here.
3. Removes the PathogenicGeneVariantData struct. I think this was confusing, and unnecessary.

Still to do:

1. Implement the methods on GIBuilder
2. Fix the tests

[SmartMonkey-git]

I think, this is a simpler, but im unsure if it does the same.

Suggested change

	let (allele_one_col, allele_two_col) = match linked_hgvs_cols.len() {
	0 => (None, None),
	1 => (Some(linked_hgvs_cols[0]), None),
	2 => (Some(linked_hgvs_cols[0]), Some(linked_hgvs_cols[1])),
	n => {
	return Err(CollectorError::ExpectedAtMostNLinkedColumnWithContexts {
	table_name: patient_cdf.context().name().to_string(),
	bb_id: bb_id.to_string(),
	contexts: vec![Context::Hgvs],
	n_found: n,
	n_expected: 2,
	});
	}
	};
	if linked_hgvs_cols.len() > 2 {
	return Err(CollectorError::ExpectedAtMostNLinkedColumnWithContexts {
	table_name: patient_cdf.context().name().to_string(),
	bb_id: bb_id.to_string(),
	contexts: vec![Context::Hgvs],
	n_found: linked_hgvs_cols.len(),
	n_expected: 2,
	});
	}
	let (allele_one_col, allele_two_col) = (
	linked_hgvs_cols.first().copied(),
	linked_hgvs_cols.last().copied(),
	);

[psnairne]

Your version will return two of the same column when len = 1 which isn't what we want (we want (col, None)).

But I've done something similar now with

        let allele_one_col = linked_hgvs_cols.get(0).copied();
        let allele_two_col = linked_hgvs_cols.get(1).copied();

[SmartMonkey-git]

I'm not sure, I understand this. You are picking a single row and then you always pick row 1 and 2.
This does not seem robust. What if there are three values and the first one is empty?

[psnairne]

I don't understand what you mean.

What happens is this: we allow one linked gene column, and up to two linked HGVS columns. Then we go row by row and collect

gene = Option<&str>, hgvs1 = Option<&str>, hgvs2 = Option<&str>e

It's almost exactly the same as how we do collection for phenotypes and diseases etc. The only difference is there can be between 0 and 2 HGVS columns