feat(grouping): Add new grouping test inputs

[lobsterkatie]

This adds a number of new grouping variants inputs:

• Two built-in fingerprint/chunkload error inputs: a base one and one with a hybrid fingerprint, to show that (in theory), the one with the hybrid fingerprint comes out with a different hash than the regular one. (I say in theory, because right now we have a bug in the system such that built-in fingerprints clobber the added part of a hybrid fingerprint. So the snapshots for these inputs do not in fact have different hashes, but once that bug is fixed, we'll be able to see that it is because the hybrid fingerprint version of the pair will have its hash change.)

• Two custom fingerprint inputs to go along with the existing, client-fingerprint-only one, which has been renamed to reflect its relation to the two new ones: a server-side-rule-only input and an input which has both a client fingerprint and a server-side rule. These are again to illustrate what should theoretically be happening (but currently isn't), which is that the client and server-side fingerprints both work, and in the case where both exist, server-side fingerprinting takes precedence. As it stands, the client fingerprint works, the server-side one is totally ignored, and in the case with both, it's the client rather than the server-side fingerprint that wins (again because we're ignoring the server-side fingerprint) - but once this is fixed, we'll similarly have proof that it has been.

• Five hybrid fingerprint inputs:

  • A set of three consisting of a base input, one that has the same added fingerprint value but different underlying data, and one that has the same underlying data but a different added fingerprint value. All three have different hashes, indicating that both the {{ default }} and the added parts are being taken into consideration.
  • A pair of inputs where there are both client fingerprints and server fingerprint rules, and in each case one or the other is a hybrid fingerprint. For the same reasons as with the previous fingerprints-from-both-sources example, when the hybrid comes from the client, it's used, but when it comes from the server, it isn't. This is another place we'll be able to see that bug getting fixed.

• A message-based-grouping input, with a message including numbers, to show that though the legacy config doesn't parameterize, the newstyle configs do.

It also adds a built-in fingerprint chunkload error input for use with the fingerprinting tests.

[JoshFerge]

nice!