Thank you Naib! That is the answer I was looking for ... It seems that an app can be written to find the log files on the solo and add the geocoding ... The app would have to run on a/the PC once the files are uploaded ... The underlying question is 'why this has been done', and you answered the 'why' ...
That would be a fun project to write if I were younger and had the energy ...
TFPJ
There are a number of solutions for the geotagging currently. You have Mission Planner (via the godmode menu, CTRL+F), GeoSetter (my favorite), and through an add-on purchase within Solex (I have no experience with this quite yet).
There are a number of changes happening in upstream ArduPilot/ArduCopter which will hopefully improve the accuracy of the geotagging results regardless of which method you choose.
1) A number of users reported poor/variable logging and timestamping, and this has been iterated upon in AP, which should result in more deterministic differences between logged time and "triggered" time. I don't have the PRs on hand for this, but if I recall correctly, this issue is being addressed from about 3 different angles. Final result should be that logged time is deterministically offset from actual time by about 40ms, and trigger events reflect this, so timestamps attached to events are actually the timestamps that should be attached, not something about 20-400ms in the past (currently).
2) Matt has written some updates to OpenSolo that will force a GoPro clock sync to the GPS on the Solo at boot/arm. This should help with another huge source of error in the geotagging process, the systematic offset of timestamps between the GPS/log data and the time the GoPro is synced/set to. I've asked about triggering this sync prior to each photo and was informed that:
A) it might not be feasible time-wise depending upon photo interval
B) GoPro clock doesn't drift meaningfully during 10-20min periods while powered on
C) GoPro clock has an internal resolution of 1s, so there is little benefit to constantly re-syncing to the GPS since it is so coarse
So, hopefully soon, even if the process doesn't necessarily change, our results should be far more robust. I'm super excited.