Don't store data filename in corpus data structure

[johnml1135]

A fix for #547.

Tests have not been updated yet - wanting to get review of approach.

---

This change is 

[ddaspit]

Reviewed 31 of 31 files at r1, all commit messages.
Reviewable status: all files reviewed, 5 unresolved discussions (waiting on @johnml1135)

---

src/Serval/src/Serval.Translation/Services/EngineService.cs line 525 at r1 (raw file):

    }

    public Task DeleteAllCorpusFilesAsync(string dataFileId, CancellationToken cancellationToken = default)

I noticed that this method was never updated to reflect the new corpus model changes. We should fix this.

---

src/Serval/src/Serval.Shared/Contracts/CorpusResult.cs line 8 at r1 (raw file):

    public required string Language { get; init; }
    public string? Name { get; init; }
    public required IReadOnlyList<CorpusFile> Files { get; set; }

We don't want to pass models between contexts. We should only be passing contracts. A CorpusFileResult should be used to pass file information.

---

src/Serval/src/Serval.DataFiles/Models/CorpusFile.cs line 5 at r1 (raw file):

public record CorpusFile
{
    public required DataFileReference FileReference { get; init; }

I would prefer not to denormalize the database within a context. We shouldn't keep copies of the data. We also don't want clients to have to specify the same information twice. This should just be a FileRef property.

---

src/Serval/src/Serval.Shared/Models/CorpusFile.cs line 9 at r1 (raw file):

    public required string TextId { get; set; }

    private string? _filename;

I see where you are going by not storing the filename and then synchronously calling Serval.DataFiles to get the filename when it is used. Rather than redesigning all of this, I would prefer that we stick with the current design. We are already dealing with one type of data file change, specifically deletion. We should just use the same approach when a file is updated. I would also like to keep the number of synchronous calls across contexts down to a minimum. In the current design, we only make these calls on setup. I would prefer not to make them on more frequent calls, such as starting a build or generating USFM. See the DataFileDeletedConsumer for an example.

---

src/Serval/src/Serval.Shared/Models/MonolingualCorpus.cs line 3 at r1 (raw file):

namespace Serval.Shared.Models;

public record MonolingualCorpus

This should be moved back to Serval.Translation.

[johnml1135]

Let me see if I understand the basic changes you are suggesting:

• Make a DataFileUpdated consumer - that updates:
  • The Corpus (as in the corpus controller)
  • The Legacy corpus
  • The Parallel Corpus
• Make a CorpusFileUpdated consumer that updates the Parallel Corpus (when the files change)

[johnml1135]

Start by writing integration tests (that fail) so that the logic can be more easily debugged.
Make "removeAll" work with parallel Corpus.
Ignore Assessment Engine.