Remote application monitoring

We are researching the possibility of adding monitoring to our applications using PRTG. We have several enterprise applications that run on central server farm and have several remote locations with many workstations running our client.

We have a couple thoughts:

1. Have a remote probe at each location (Group) and setup each workstation (Device) with a single HTTP Data Advanced sensor that calls into our application for monitoring information and returns multiple channels for different statistics.
2. Have a remote probe at each location (Group) and setup each workstation (Device) with multiple HTTP Data Advanced sensors that call into our application for monitoring information for different statistics.
3. Utilize the Mini Probe API and have our application send the monitoring information.

A couple questions:

1. Is the limit on Mini Probes really "Less then 100 probes" per installation (Documentation)?
2. Does the Mini probe call to a remote probe or central? If remote, does the remote batch information and send to central?

Is there a suggested way to accomplish this goal?

http-sensor mini-probe prtg remote-probe