As Don says, this game is not a form of "military chess" like some wargames can be. You can figure out that if you do A and then B, the result
should be C
most of the time, but result C is not 100% guaranteed.
Morale tests get taken for various things, including (most probably in this case) engaging enemy at 1 hex or less ("firer panic" message if spooked).
In addition there will be a generic situational morale test at end turn. This takes things like suppression, isolation from friends, how many casualties your platoon has, how many of your platoon mates are retreating or worse, whether you are in contact with a higher leader, whether you are a tank in woods with enemy grunts nearby and all these sorts of situational modifiers into account. So at end turn, some units will so well on morale test and deplete suppression, most will stay as-is, and some will get spooked.
Unlike in say ASL where you had to go manually trough game phase 2.3.4.5 End Turn Morale Test, then apply rule 1.2.3.4 Tank isolation and rule 4.5.2.1 situation 1.2.3 (Commander opens a "Dear John letter"
) the morale tests are all done in the code and the results are shown as they fall (e.g. some unit decides to run away).
The end of turn morale check could be what this one failed. Difficult to tell really since the save game is the usual sort of save one gets from a "bug complaint" i.e.
the save is post event so one cannot repeat the action the user performed trace the code via the debugger, and perhaps only run through the game's replay function - which does not log all messages, just key events like who is firing at what.
A post event save merely shows that an effect occurred. For debugging, one wants a save from
just before it happened, and information from the end user on what
precisely was done to make the event happen. (Then of course the event wont happen when running in the debugger, as it was a once in a blue moon which was also a Thursday type of thing
).