PingPlotter sends packets continuously. To stop it at 100, change "# of times to trace" to 100. The tech probably wants 1 ping per second, so change "Trace Interval" to 1 second. Also, since you want to look at all your samples, you can change "Samples to include" to 0 (ALL) or 100. Because you're only taking 100 samples, these will be equivalent. If you want to send samples more quickly than one per second, you can decrease the trace interval (enter a smaller number - like 0.5 seconds). 1 second is a good setting, though, and usually results in useful statistics.
The "Err" column is the count of lost packets. You can also term this as "lost". If you collect 100 samples, then the Err column should match the PL% column. As you collect more or less samples, though, these will diverge.
In reality, I'd probably leave "# of times to trace" set to 0 (unlimited) and set "Samples to include" to 100. This will mean any pictures you create, or text you copy (from Edit -> Copy as Text) will be for 100 samples, but you can collect data continuously and focus on the 100 samples that most closely represent the problem you're trying to communicate to Rackspace. You can focus the upper graph statistics on a non-current period by double-clicking on the time graph for the period you want to look at. (This only works if you have more samples collected than your "Samples to include" setting is, though - if you have 1000 samples collected and only want to look at 100 (samples to include=100), then it works great. If you only have 10 samples collected, though, it doesn't do much good since it's hard to find a subset of 100 packets when your dataset size is only 10 packets).