allow the same duplicate/merge to be suggested multiple times

[chris48s]

Refs https://app.asana.com/1/1204880536137786/project/1204880927741389/task/1214041815638938?focus=true

This PR:

• Allows more than one DuplicateSuggestion for the same pair of people. You still can't have a duplicate if one is open
• Surfaces any previous rejections in the DuplicatePersonView (suggest a duplicate) and DuplicateSuggestionListView (review duplicates)
• Update + add tests

Example output:

This PR is mostly Claude with a bit of my own guidance/judgement/review.

[chris48s]

I was in two minds about whether this should assume the pairs are ordered (low, high) or accept them in any order.
I decided to go with assuming they're in (low, high) order because it makes it simpler what the function should return.

To work an example, lets say there is a DuplicateSuggestion record DuplicateSuggestion(person=1, other_person=2)

The way this works now, get_previous_rejections([(1,2)]) returns {(1,2): [DuplicateSuggestion(person=1, other_person=2)]} and get_previous_rejections([(2,1)]) returns {}

If I make it so that get_previous_rejections([(2,1)]) matches that record too it is ambiguous to me whether the output should be:

{
    (1,2): [DuplicateSuggestion(person=1, other_person=2)]
}

or

{
    (2,1): [DuplicateSuggestion(person=1, other_person=2)]
}

or

{
    (1,2): [DuplicateSuggestion(person=1, other_person=2)],
    (2,1): [DuplicateSuggestion(person=1, other_person=2)],
}

I feel like saying "the inputs have to be ordered and so is the output" just makes things simpler.

I could change it though, if we want.

[symroe]

Yeah, we decided to always go with ordered in the rest of this feature, so let's stick with that here. I think it also makes it clearer that high IDs always merge down to low IDs