Change "This buffer was empty." tooltip text to "No samples in this range."

[mstange]

Profile: https://share.firefox.dev/3ONUsYd

If you move your mouse over the striped area an the start of the track, the tooltip says "This buffer was empty.".

The sentence "This buffer was empty." is pretty much always inaccurate. It's supposed to mean "We probably discarded some data here because the buffer was full.". But these days the striped area is pretty much only shown in two cases:

1. When using the Gecko profiler on Linux, where we miss data near the start of a new process because of unwinder initialization.
2. When using other profilers which don't sample all threads at the same rate. For example when using the perf importer, Chrome importer, or other importers.

I think "No samples in this range." would be true in more circumstances.

Background: When we added this tooltip, the Gecko profiler had one buffer per process, and the content process would often run out of buffer space before the parent process ran out of buffer space. This created an imbalance at the beginning of the profile, where the parent process had samples but the content process did not. These days, the pool of buffer chunks is shared across all processes, the maximum buffer size is much higher, and when we do run out of buffer chunks, all processes discard data around the same time. So this imbalance no longer happens.

┆Issue is synchronized with this Jira Task

[fqueze]

I see the "ran out of buffer space" case often when capturing long profiles. It happens because some long markers are recorded with a duration when they end, so they are stored in recent buffer chunks, but the buffer chunks related to when the markers started have already been recycled.