I wonder if the LM1085's used at the TRIAC locations were a manufacturing error that didn't get caught during their 'strenuous testing'.
Regarding TRIACSs --
MAC210A8G -- formerly made by On-Semiconductor (Motorola). On-Semi sold off TRIACs and SCR's to Littelfuse (yes, the fuse manufacturer) so they are now made by Littelfuse.
Just a quick scan through the data sheets and compared the MAC210A8G to the often used BT138-600E:
Both are rated for 600V which is more than adequate.
MAC210A8G has a 10 amp rating, BT138 has a 12 amp rating.
The MAC210 requires 3x to 5x the amount of gate current (IGT) to trigger the TRIAC.
The MAC210 has a slightly higher VGT trigger voltage than the BT138.
Since the gate is driven by a 2N5401 - I don't see VGT or IGT being the issue.
Not a lot of difference but the BT138's were a tiny bit better than the MAC210s in nearly all other specs, I didn't see any real show stoppers other than the three small ones listed above.
Note - I only glanced through the specs and didn't do any real digging into this.
One thing that can make a big difference -- is the heatsink used by RD the same as those used on the original Wms boards?
Probably the reason they used the MAC210s - they're widely available while the BT138-600E parts are a bit harder to find and cost more.
The pinouts were the same and so is the biasing circuit - the BT138-600E parts can be a direct replacement.