Improve mnist tutorial (#147)

* Minor: Removed a divide by zero warning from the fixed point LIF ProcessModel

Signed-off-by: Risbud, Sumedh <sumedh.risbud@intel.com>

* Improved to MNIST end-to-end tutorial
    - uses fixed point bit-accurate ProcessModels for LIF and Dense
    - resets internal neural state of all LIF neurons
    - these changes are needed to make the pre-trained networks parameters work, because the network was trained with these assumptions

Signed-off-by: Risbud, Sumedh <sumedh.risbud@intel.com>

* Post code review @awintel and @phstratmann

Signed-off-by: Risbud, Sumedh <sumedh.risbud@intel.com>

* Post code review @awintel and @phstratmann

Signed-off-by: Risbud, Sumedh <sumedh.risbud@intel.com>

* Post re-review by @phstratmann

Signed-off-by: Risbud, Sumedh <sumedh.risbud@intel.com>

Co-authored-by: PhilippPlank <32519998+PhilippPlank@users.noreply.github.com>