fix(tts): guard SpeechSynthesisUtterance addEventListener calls

[sentry]

This PR addresses a TypeError: r.addEventListener is not a function occurring in use-tts.ts when SpeechSynthesisUtterance.addEventListener is called unconditionally.

The root cause is that in some environments (e.g., older Safari, certain Android WebViews), SpeechSynthesisUtterance might not expose addEventListener as it doesn't extend EventTarget in those specific contexts. Calling an undefined method results in the observed TypeError.

The fix involves adding a typeof utterance.addEventListener === "function" check before attempting to use addEventListener. If addEventListener is available, it is used as before. Otherwise, the code falls back to assigning event handlers directly to the onstart, onend, onpause, and onresume properties of the SpeechSynthesisUtterance object. This approach aligns with existing defensive patterns in the codebase, such as those found in speech.ts for window.speechSynthesis.

Fixes ECENCY-NEXT-1GNS

_{This PR was automatically generated by Sentry. You can adjust this setting at any time.}