Fix incorrect usage of Timer.Reset, honour Context deadlines, avoid leaking timers

[cPu1]

What?

TODO: open this PR in the upstream repo after reviewing the contribution guidelines.

I came across this while going through the codebase.

This changelist includes three changes:

• Fix incorrect usage of Timer.Reset.
• Honour Context deadlines, including for http.Request.
• Avoid leaking timers.

There are a few places in the codebase where Timer.Reset is being called right before attempting to receive from the
Timer.C channel. If the timer has already fired, this can lead to a receive from Timer.C returning earlier than
the duration supplied to Timer.Reset as Reset does not drain the Timer.C channel.

Here's an example to demonstrate this:

timer := time.NewTimer(100 * time.Millisecond)
time.Sleep(200 * time.Millisecond)
timer.Reset(24 * time.Hour)
select {
case <-time.After(1 * time.Millisecond):
case <-timer.C:
  panic("fired before 24 hours")
}

It's not advisable to call Timer.Reset while it's still active on a Timer obtained from a call to time.NewTimer,
as mentioned in the docs. The correct usage is to call Reset only on timers
that are stopped or have had their channels drained.

This changelist fixes the issue by creating a new Timer before waiting on Timer.C, and Stopping it before the function returns.

Why?

Checklist

• I have performed a self-review of my code.
• I have added tests for my changes.
• I have run linter locally (make lint) and all checks pass.
• I have run tests locally (make tests) and all tests pass.
• I have commented on my code, particularly in hard-to-understand areas.

Related PR(s)/Issue(s)