GitLab

- Added preliminary connection classes with neural network class able to evaluate inputs via the feed-forward method.
- Added a test case for validating feed-forward capabilities of our networks

preliminary work on connections, neural networks and neurons with neural networks as activation functions

to connection_m, net_m and neuron_m to make serializing instances of connection_t, net_t and neuron_t
possible despite having allocatable and pointer components. NOT WORKING YET!

Partly implemented method new_mock_net() in the module net_m. NOT FULLY DONE YET!

TMP (uncompilable): Attempting to use basic data types from module data_kinds_4neuro_m + attempting to solve compilation problem

* Added missing 'return' statements to constructor in Doxygen comments
* 'sigmoid' neuron renamed to 'logistic'
* Logistic neurons' activation function changed to generalised logistic
function (added new constructor with 'alpha coefficient' parameter)
* Removed out-commented blocks of code with cancelled interconnection_t
dependencies

ENH: Implemented classes mock_neuron_t, binary_neuron_t, linear_neuron_t and sigmoidal_neuron_t in the module neuron_m.