AverageBear: 3. For units based on activity
If no other unit takes any action, ideally the Wizard should get full from 1 & 3 above. Plus given low number of units, also expected to get a lot from 2.
I am comparing this with Scout which often ends up in a similar situation especially post level 10 with double shot. Please correct me if I am wrong...
UnitActRating = 20*MeleeKills + 10*RangedKills + 2*MeleeDamage + RangedDamage + DamageReceived + 3*SpellsStamina + 2*UsedHealing + UsedStamina + 1
I suspect that main idea behind ditching out spell damage along with poison from exp distribution formula is potential issues with tracking source and amount of damage. Like mage would cast mass disease then ghoul would hit diseased target and apply disease again. Or problem with 20 poison damage on target with 5 hp left.
So, for activity you will get only spellstamina and stamina.
If you won by using fireball on first turn, you would have activity rating as :3*4(spellstamina)+4(total stamina)+1=17.
Compare it with scout which won against, say 2 minotaurs with 2 doubleshots:
10*2(ranged kills)+104(damage)+ 10(stamina) +1 =135
If they both have 5 units then mage would get 17/22=77% of activity exp, scout 135/140=96%.
Thing is, on the way to these battles you pretty often will have situations where swordsman would do 2-3 melee kills and receive more activity rating than wizard which spent all his spellbook on damage spells.
Buffer/debuffer wizard, on the other hand, can disable enemy by web/fatigue/blind, buff himself and score meleekills.