feat!: Add `EvaluationContext` helpers and context merging to flag evaluation

[maxveldink]

This PR

This is largely a port of this behavior from the Sorbet implementation my company originally started with.

• Updated arguments to be evaluation_context across the board more explicitly
• Ensures that we allow for API, Client and invocation contexts to be passed
• Refactor EvaluationContext to have a cleaner initializer signature and add == and merge convenience methods
• Use EvaluationContextBuilder during flag evaluation to reconcile all contexts
• Removes test that is now covered in specification tests

Related Issues

Closes #111

Follow-up Tasks

• Still need to implement transaction propagation for evaluation contexts
• More tests for client flag evaluation

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 99.46%. Comparing base (6895a0d) to head (984e60a).

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #119 +/- ## ========================================== + Coverage 99.42% 99.46% +0.03% ========================================== Files 14 15 +1 Lines 174 187 +13 ========================================== + Hits 173 186 +13 Misses 1 1 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[maxveldink]

@beeme1mr Good callout, added some examples

[maxveldink]

I don't think we're overly concerned with thread safety here. We haven't run into any issues with a similar implementation yet. Happy to take a look at any bugs that pop up around that though!