This one's a bit difficult to describe, but it's a problem with how the engine handles opcode 1, attacks per round bonus. From the IESDP (emphasis mine):
#1 (0x001) Stat: Attacks Per Round Modifier 
Parameter #1: Key Modifier
Parameter #2: Type
Alters a characters Attacks per Round, by changing the Key by the modifier value specified by the 'Key Value' field, in the style specified by 'Type' field.
Key Attacks Per Round
Known values for 'Type' are:
0 Cumulative Modifier -> Key = Key + 'Key Modifier' value
1 Flat Value Modifier -> Key = 'Key Modifier' value
2 Percentage Modifier -> Key = (Key * 'Key Modifier' value) / 100
3 Cumulative Modifier -> Same as 0NB. When this opcode is stacked, the values of the Key Modifier are stacked, not the number of attacks.
The problem comes with effects stacking. Since it's the key value that's being added, a bunch of effects can push the value from 5 to 6 resulting in a drop from 5 attacks per round to 1/2 attack--exactly what happens with Melf's Minute Meteors gets crossed with Boon of Lathander. Melf's sets attacks to five (sets key=5) and then Boon gives a bonus of one attack per round (key+1), resulting in a Boon'd, MMM-wielding cleric-mage to throw a meteor every other round (key=6). We worked around this in BG2FP by disallowing casting MMM when under the influence of Boon or Offensive Spin, but it's a nonsensical hack
edit: BG2FP worked around this by splitting the ApR bits of Boon/OS into a separate spells and blocking those spells when wielding MMM, which is wonky but functional.
My question is whether we can modify the opcode to behave more sensibly. In hindsight the key value would have been laid out linearly so that 1 = 0.5apr, 2 = 1apr, 3 = 1.5apr ... 10+ = 5apr and spells that added a full attack per round would just go key+2. It'd be trivial to adjust any BG/BG2 spells to use the new convention, however, this would break all sorts of mod stuff.
At the least would it be possible to prevent the key value from crossing over from 5 to 6?