Timing information in temperature measurement

[JorisVincent]

Currently, temperature measurements are timestamped, but the timestamp is in mglGetSecs, which is relative to computer startup time. This is not of much practical use.

Either we change this to second-accurate datetime, or we get rid of it. Especially when temperature is part of a validation, the timestamp is currently cluttering a struct, in a struct, in a struct, in an object. Moreover, datetime of validation is already recorded in the validation struct.

[npcottaris]

The temperature timestamps are obtained via mglGetSecs because so are the timestamps of spd measurements. If you change it, you probably need to change both so you can relate temps to spds.