Reland "afp: Attempt to retrieve the ThrottleManager from NavigationHandle"

This is a reland of commit 1dbd686ae1e4f11e289c8e960e793880ee23f3aa

The original CL was reverted
[here](https://chromium-review.googlesource.com/c/chromium/src/+/6177875)
because of a suspected memory regression. The regression seen in the
benchmark is not present during finch experiments, and the change seems
to make the benchmark more deterministic. See bug for more details.

> afp: Attempt to retrieve the ThrottleManager from NavigationHandle
Original change's description:
> During a navigation where the RenderFrameHost is being re-used, calling
>
> `RenderFrameHost::GetPage()` could return the Page from the old origin.
> For example, when navigating from `about:blank` to `a.example`,
> `RenderFrameHost::GetPage()` might return `about:blank`'s Page.
> This causes us to retrieve the wrong ThrottleManager which results in
>
> the wrong ActivationState being sent to the renderer.
> To solve this, first try to retrieve the ThrottleManager from the
>
> NavigationHandle.
> Bug: 382645158
>
> Change-Id: I2fd677456c21d031741cb47ea542718efb53b67e
> Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/6107451
> Commit-Queue: Giovanni Ortuno Urquidi <ortuno@chromium.org>
> Reviewed-by: Zainab Rizvi <rizvis@google.com>
> Cr-Commit-Position: refs/heads/main@{#1404397}

Bug: 382645158,388884058
Change-Id: I60601ffc50a90ee3f09a41345675decd9015d246
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/6187562
Reviewed-by: Zainab Rizvi <rizvis@google.com>
Commit-Queue: Giovanni Ortuno Urquidi <ortuno@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1410693}