Wrong behavior at the deadlock when systems cannot expose quantities

Before, setting InteractionTiming = timig_exact was causing a deadlock at the first iteration, producing no output to the stdout or to the system logs. This behavior was not the appropriate one, as the code should stop with a message that the propagation is not taking place due to the deadlock. But now, the stdout is writing the iteration steps and times as if the calculation was running (!), which is even worse (although the system logs still don't show any output even at the initial timestep).

We need to fix this behavior by:

• Stopping the code safely when the deadlock is found.
• Adding some tests for the wrong combination of timesteps + interaction timing.
• Change the default InteractionTiming to retarded, as it is a safer default, with proper warnings.
• The Debug = info mode no longer provides useful information for this deadlock.

@micael.oliveira please take a look at this.

Input files that reproduce the problem:

inp_gs

inp_td

12-tddft-currents-to-maxwell.00-benzene.xyz