MCR-4097 Fix frontend monitoring, add back error boundary fallback UI

[haworku]

Summary

• Get proper error component rendering again when JS error thrown
  • See 4910df5
  • Fixed this error by bringing back <PageHeading/>. This is needed because theusePage context is not reliable if the entire app is crashing, it's too far down in the stack and will not be available in a crash.
• Get thrown errors logging again in monitoring (added New Relic snippet to handle this)
• New Relic config set up in higher level environment

Config changes
I added 4 new config for each environment for new relic. Several are the same across all environments. I think eventually REACT_APP_NR_LICENSE_KEY could totally replace existing NR_LICENSE_KEY used to set up OTEL - but decided that could switch on future PR since its a change to existing setup

New Relic Changes

• set up three new applications for NR browser monitoring mc-review-web (for dev), mc-review-web-val, and mc-review-web-prod
• added errors from val and prod to the app-web errors alert policies
• added to dashboards

Related issues

https://jiraent.cms.gov/browse/MCR-4097

Test cases covered

No tests but made a way to manually check monitoring

QA guidance

• Login in as Admin user
• Navigate to /settings?test=crash

You should see

1. the error boundary which is a react error component NOT a blank screen
2. an error in the browsering monitoring for that environment alert in slack if this is triggered in a higher environment
   
3. (if this is set up) an alert in mc-review alerts channel on slack

[mojotalantikite]

STAGE_NAME in dev is confusingly main, so we'll want that in this array. Really not sure why it was called main 🤷🏻‍♂️