As I understand it, the criteria for awarding the LM points is a successful "LEM THRUST TEST" step, which should be an 'a' code in the mStr. When Mev[i].PComp is set in MissionSteps, there is no case 'a' defined. This makes sense; a thrust test isn't a prestige event. Consequently the thrust test should fall under

 399     default:
 400         Mev[step].Prest = -100;
 401         Mev[step].PComp = 0;
 402         break;
 403     }

and PComp should remain at that value for the rest of the function (the steps fails the subsequent if... conditions)

 405     if (Mev[step].Prest != -100) {
 406         Mev[step].PComp = WhichPart(plr, Mev[step].Prest);
 407     }
 408 
 409     if (PastBANG == 1 && Mev[step].PComp > 0) {
 410         Mev[step].PComp = 5;
 411     }

The PComp value isn't referenced again until FailEval in mis_m.cpp; for a LEM Thrust Test, the failure codes should be 0 (A-OK), 31 (everyone dies), or 4 (fail and branch to alternate). Every instance where PComp is modified in FailEval - in fact, in all of mis_m.cpp - the code always checks

if (Mev[STEP].PComp > 0) {
    Mev[STEP].PComp = 4;
}

Since PComp should be 0 at this point, it will remain that value.

When the steps are traced in prest.cpp, Mev[i].PComp once again may be changed

 804     other = MaxFail();
 805 
 806     for (i = 0; i < STEPnum; i++) {
 807         if (Mev[i].PComp == 5 && Mev[i].StepInfo == 0) {
 808             Mev[i].PComp = 0;
 809             Mev[i].Prest = -100;
 810         }
 811 
 812         if ((MANNED[0] + MANNED[1]) > 0) {
 813             if (other >= 3000) {
 814                 Mev[i].PComp = 4;
 815             } else if (Mev[i].Prest >= -28 && Mev[i].StepInfo > 2999) {
 816                 Mev[i].PComp = 4;
 817             }
 818         }
 819     }

and should be set to 4 if somebody died. For a type 31 failure, this is true. For a type 4 failure, it need not be. The max fail code could easily be 1926, the value set for a type 4 failure during the LEM Thrust Test. In that case, Mev[i].PComp would remain 0 and award LM points in Set_LM even though the step failed.

Am I missing something?

You are right, I wasn't aware of the check in MissionSteps. Then we should add a test for StepInfo == 50, such that a-ok failures also get a pass. I have updated the commit to do that.