Following up on T415708: Synthetic experiment to test new event path and T415790#11604075:

RECOMMENDATION: An experiment where we test both external paths using navigator.sendBeacon() directly AND also use mw.tk.getInstrument().submitInteraction() to evaluate how much event loss all the stuff in the middle is responsible for.

This work builds on externalPathTest.js (instrument config, experiment config)

We will keep the configurations mostly the same and only need to adjust the start & end dates of both the experiment and the instrument.

Acceptance criteria

• In addition to the current methods for sending events, the event is sent to the new external path (/ins-502b/v2/events) directly via navigator.sendBeacon(), for 3 events total per page visit for users enrolled in the experiment
• All 3 events include UA strings
  • synth-test-external-path configured to collect agent_ua_string contextual attribute
  • Events sent via sendBeacon() (directly) manually populate agent.ua_string property in the event data
• action_context includes JSON-stringified object with properties:
  • tk_sdk (boolean) true for event sent via submitInteraction() and false for other two events sent via sendBeacon directly
  • ext_path (string) "old" or "new"
  • tz_offset (int) output of getTimezoneOffset(), which we will use to study how the class of device used varies by local (not UTC) time of day
• Data collection is active for 7 days