Remove deprecated non-parallel corpus from Serval

This change would constitute a major breaking change. This work would be a continuation of https://github.com/sillsdev/serval/pull/497 and https://github.com/sillsdev/serval/pull/535.

When we do this, we should also properly address the DTO naming collision issue across translation and word alignment. (From Damien's comment here) This impacts AlignedWordPairDto, ParallelCorpusFilterConfigDto, ParallelCorpusFilterDto, TrainingCorpusConfigDto, and TrainingCorpusDto. The name collisions happen in the Swagger JSON file as well. This means that we need to make a breaking change to get this right. We have two options:

• Prefix all DTO names with "Translation" or "WordAlignment". This was what I originally was trying to do, but clearly failed, since I allowed some DTO names without the prefix. We can do this automatically using the ServalSchemaNameGenerator. We could remove the prefix from DTOs that already have the prefix, i.e. TranslationEngineDto could be renamed to EngineDto.
• Add custom hooks in NSwag to generate DTOs in "Serval.Client" with different namespaces. For example, Serval.Translation.AlignedWordPairDto would be generated in the client as Serval.Translation.AlignedWordPair instead of Serval.Client.AlignedWordPair. The translation engines client interface would be generated as Serval.Translation.ITranslationEnginesClient. I don't think NSwag supports this out-of-the-box. We would need to create our own console application that calls NSwag. See [this](https://github.com/RicoSuter/NSwag/issues/2728 for more information).

Start by reviewing the linked pull requests, the ServalSchemaNameGenerator, and the DTOs named in the issue, then investigate how NSwag produces the Swagger JSON and Serval.Client types. Compare the two proposed naming approaches; done means the deprecated non-parallel corpus is removed and the translation and word-alignment DTO names no longer collide.