Clean up legacy build error and build failure models

[williamjallen]

Incremental progress towards the eventual combination of these models. The last significant difference between these models is the build failure arguments, which I plan to address in a follow-up PR.

[josephsnyder]

I've got some errors on attempting to load Build Error and Build warnings in both new and old data. Both are 500 responses:

• str_contains(): Argument #1 ($haystack) must be of type string, null given
• basename(): Argument #1 ($path) must be of type string, null given

Each prevents the load of information into each page
Using testing Trilinos Data:

Using newly submitted build failures:

[williamjallen]

@josephsnyder It looks like that error was caused by reviewing #3432 and then switching to this branch (the same thing that caused conflicts). I just rebased, and the error should be resolved.