chore: refactor navigation

[sadym-chromium]

Addressing Align navigation with the spec.

Navigation tracker relies on the following events:

• Page.frameNavigated
• TargetEvents.FrameStartedNavigating (Network.requestWillBeSent)
• Page.navigatedWithinDocument
• Page.frameRequestedNavigation
• Page.javascriptDialogOpening:beforeunload

BEGIN_COMMIT_OVERRIDE
fix!: align navigation started with the spec
END_COMMIT_OVERRIDE

[OrKoN]

There are a few new Puppeteer failures as well as far as I can see.

[OrKoN]

The following Puppeteer tests are new regressions with this PR:

• "[navigation.spec] navigation Page.goto should work when reload causes history API in beforeunload" FAIL
• "[navigation.spec] navigation Page.goto should work when navigating to data url" FAIL
• "[navigation.spec] navigation Page.goto should navigate to dataURL and fire dataURL requests" FAIL
• "[requestinterception-experimental.spec] cooperative request interception Page.setRequestInterception should navigate to dataURL and fire dataURL requests" FAIL
• "[requestinterception.spec] request interception Page.setRequestInterception should navigate to dataURL and fire dataURL requests" FAIL
• "[oopif.spec] OOPIF should load oopif iframes with subresources and request interception" FAIL

[OrKoN]

I have landed the roll with existing changes into Puppeteer so that new regressions are clear. If anything fails in the Puppeteer job now, it is caused by the changes in this PR.

[OrKoN]

I have landed the roll with existing changes into Puppeteer so that new regressions are clear. If anything fails in the Puppeteer job now, it is caused by the changes in this PR.

I'd suggest we add e2e tests/fixes for Puppeteer scenarios that this change affects.

[OrKoN]

"[oopif.spec] OOPIF should load oopif iframes with subresources and request interception" seems to be a Linux-only failure upstream.