refactor: remove Types collection from wdl-analysis.

[peterhuene]

This commit removes the Types collection from wdl-analysis.

The collection served as an arena to allocate types into which also served as providing type recursion via arena identifiers. The hope was that this would prevent a lot of small allocations from occurring during analysis when calculating expression types.

However, each document had a different Types collection, requiring "importing" types between collections. Further complicating things, an evaluation engine had its own Types collection which needed importing of types from other document collections. This was a source of bugs in the implementation of the evaluation engine.

Worse, the types collection served as a bottleneck for sharing data across asynchronous tasks in the upcoming implementation of workflow evaluation, requiring a lock.

The alternative is to use an Arc in appropriate places where type recursion occurs. After running some simple benchmarks, the removal of the types arena doesn't appear to have any real world impact on performance.

Additionally, the removal of Types allows Type to implement PartialEq rather than a custom trait that uses the arena; likewise for Display.

Similarly in wdl-engine, Value can now directly implement serialization and deserialization traits.

Before submitting this PR, please make sure:

• You have added a few sentences describing the PR here.
• You have added yourself or the appropriate individual as the assignee.
• You have added at least one relevant code reviewer to the PR.
• Your code builds clean without any errors or warnings.
• You have added tests (when appropriate).
• You have updated the README or other documentation to account for these
  changes (when appropriate).
• You have added an entry to the relevant CHANGELOG.md (see
  "keep a changelog" for more information).
• Your commit messages follow the conventional commit style.

[peterhuene]

No rush on this review as I'm out this week and next, but the workflow evaluation implementation will be relying on these changes so that we can more easily spawn asynchronous tasks that do WDL evaluation internally (i.e. we no longer have to synchronize a singular Types collection for things to work; plus, it just makes the wdl-analysis and wdl-engine implementations cleaner).

[peterhuene]

Looks like gauntlet caught a bug for me to track down too.

[peterhuene]

So the gauntlet diff is actually removing false positives fixed by this PR.

The actual bug was in the now-deleted import function responsible for importing a type between two different type collections (e.g. referencing an input for a task/workflow in an imported document). The bug was that the imported type accidentally dropped the optional qualifier for the type, resulting in type coercion failures in task/workflow call statements.

For example:

bar.wdl:

version 1.1

task foo {
    input {
        Array[String]? bar
    }

    command <<<>>>
}

foo.wdl:

version 1.1

import "bar.wdl"

workflow bar {
    input {
        Array[String]? bar
    }
    
    call bar.foo { input: bar = bar }
}

Results in this false positive diagnostic:

error: type mismatch: expected type `Array[String]`, but found type `Array[String]?`
   ┌─ foo.wdl:10:33
   │
10 │     call bar.foo { input: bar = bar }
   │                           ---   ^^^ this is type `Array[String]?`
   │                           │      
   │                           this expects type `Array[String]`

I'll push a bless of the gauntlet diagnostics to rid the false positives and add a test case to ensure we don't accidentally drop the optional qualifier like this in the future.

[claymcleod]

LGTM.