Reduce raciness in test

[lukedirtwalker]

fetchSegsFromDBRetry select on ctx.Done() and time.After().
If the setup/calling of ctx.Done() takes more than what we pass in tim.After,
it can be that both channels are ready at the same time and then the test might fail.

By increasing the timeout in the test this should no longer be a problem.

---

This change is 

[sgmonroy]

So just a bit more context, this is the result of investigating a CI failure of an unrelated change.
It seems that CI is very much resource constrained and race conditions show up more often.

func (h *baseHandler) fetchSegsFromDBRetry(ctx context.Context,
        params *query.Params) ([]*seg.PathSegment, error) {

        for {
                upSegs, err := h.fetchSegsFromDB(ctx, params)
                if err != nil || len(upSegs) > 0 {
                        return upSegs, err
                }
                select {
                case <-ctx.Done():
                        return nil, ctx.Err()
                case <-time.After(h.retryInt):
                        // retry
                }
        }
}

We reasoned that on each select evaluation, a timer/channel is created which spawns its own go routine. Thus, it is basically a race between the timer and the current go routines. If the current go routine does not run enough to evaluate the select cases, it is theoretically possible that the timer go routine did run and expired, which ends up in both cases being true and one of them randomly executed.

Reviewed 1 of 1 files at r1.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @kormat)

[sgmonroy]

Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @kormat)

[kormat]

Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @kormat)