Units should be consistent across neural mass blocks

[agchesebro]

We should have a discussion about this at some point - each block is internally consistent but some are on different time scales (e.g., Jansen-Rit has default parameters in seconds vs Larter-Breakspear in milliseconds). Not a huge issue now because all models are kept separate, but if you wanted to link blocks of different kinds it'll become a thorny problem.

[harisorgn]

but if you wanted to link blocks of different kinds it'll become a thorny problem.

Also a serious issue when connecting to sources that emulate some sort of trial with image presentations and pauses in milliseconds for example, and when doing reinforcement learning, updating weights every N milliseconds etc.

This package provides a nice interface for units :
https://painterqubits.github.io/Unitful.jl/stable/

[agchesebro]

This was actually mostly addressed in PR #295. Leaving this open for units, but I forgot to comment on this issue at the time. All neural masses are now in milliseconds.