[CSTP] CSTP branch has certain conditions for usage

[nmsutton]

There are some things that users should be aware of when using the CSTP branch. One of which is that COBA is on by default instead of CUBA mode as in the master branch. I think CUBA mode many not be allowed in the CSTP branch.

setConductances():
From my knowledge, setConductances() should not be set to a value because it may misdirect code to be used (not totally sure on that). Just don't include setConductances() in the code. An issue with setConductances() in the CSTP branch is that not all connections use the same tau_ds for fast and slow receptors. In the master branch, setConductances set tau_d for each receptor type. However, in the CSTP branch, tau_d is set on a connection-specific basis with the use of setSTP(). Therefore, values used in setConductances() seem not to be used because the setSTP() ones are used instead.

setSTP():
In the CSTP branch, it appears setSTP() needs to be set for any pre_synaptic_group to post_synaptic_group or synaptic signaling will not occur between them. This is also at least the 9-param (e.g., STPtdNMDA(150.0f, 0.0f), etc.) version of the function and not the master branch's 3-param version. In the master branch, COBA mode can be set to active and setSTP() not set for individual groups and those groups will still allow synaptic signaling but just not use STP (if I understand right). In contrast, in the CSTP branch, in my tests, omitting setSTP() or setting its "isSet" parameter to "false" for any connection causes synaptic signaling not to occur with that connection. I'm not sure if any other signaling is affected than synaptic signaling. In other words, every connection between neuron groups requires setSTP() to be included and set to active. Possibly this is from to tau_d being needed due to COBA being enabled by default, and without setSTP() there are no tau_ds for a connection, because the values for that from setConductances() are not used.

These things are more technically a limitation than a bug but users who are not aware of them could get different results than expected in a bug-like way. Additional code could be added to accommodate use cases where more options then these specific conditions of using CSTP code allow for. I'm not sure if @nmsutton or @jkopsick will be the ones to do that development given we may not use this code in the other use cases.