🧹 Added graceful handling of aborted requests

[ninadbstack]

What ?

• Added graceful handling of aborted requests
• Now we keep track of aborted requests and whenever we are going to make a session.send call we check if we have already received abortion event for same request id
• If request was already aborted we dont print it out as Error: Protocol error (Fetch.fulfillRequest): Invalid InterceptionId.
• We now print two new logs under debug mode, one for aborted request event and once when we ignore error above due to the same.

Why?

• There are cases when a browser triggers a network request and cancels it before its resolved. It usually happens when a resource was originally required, but is now not required [ say you have an srcset with multiple entries and you change the width of the browser fast enough such that it starts loading one of the resources, but now it doesn't need it ]
• In this case we would first receive Network.requestWillBeSent event, which we would eventually respond with Fetch.continueRequest or Fetch.fulfillRequest etc, but before we respond if the request gets cancelled [which invalidates Interception Id] we get above Protocol error.

Test:

• This was hard to write test for as its hard to reproduce this race [ or browser cancelling a request ] to begin with
• So we are using fetchWithTimeout helper so that we can mimic a cancellation from browser
• In order to reproduce the race condition we specifically delay Fetch.continueRequest response by 1 sec so that request triggered by fetch is cancelled
• Which is equivalent to calling a Fetch.* method with a request/interception Id of a request thats already cancelled

[itsjwala]

lgtm, nice!