Add a mechanism to redact fetch instrumentation

[nickdichev]

Hey guys,

I'm using better-auth in my app which uses fetch under the hood to make requests. I noticed that sensitive credentials were leaking into HyperDX via the fetch instrumentation. Obviously, I don't want this to happen.

This PR attempts to add a redaction mechanism on the fetch instrumentation. Was looking for some initial feedback on what you all think about this direction, since I think some of the other auto-instrumentation could/should use a similar pattern. I think there's a couple of other ways this feature could be implemented like modifying the HyperDXSpanProcessor or writing a custom exporter, but that would add a JSON.parse on every span (on http.request.body), which is maybe undesirable?

This work allows the application developer to instantiate their client side hyperdx like this:

HyperDX.init({
  apiKey: import.meta.env.PUBLIC_ENV__HYPERDX_API_KEY,
  service: import.meta.env.PUBLIC_ENV__HYPERDX_SERVICE_NAME || "unknown",
  tracePropagationTargets: [],
  consoleCapture: true,
  advancedNetworkCapture: true,
  instrumentations: {
    fetch: {
      propogateTraceHeaderCorsUrls: [],
      advancedNetworkCapture: () => true,
      redactKeys: {
        body: ["password"],
        headers: ["date", /content_.*/],
      },
    },
  },
});

Which will then end up like this on the console: