With the speed of the target, I'd always go for a long-range FQ shot - the aim is to approach the target from ahead rather than trying to follow it in a turn as it passes "overhead". While this means that the intercept phase is entered at a lower energy, the reduction in required turn performance, and shallower climb profile means there is a lower requirement for energy so this may come out a wash, or in favour of the slower missile at longer range.

Given the penchant for jamming, I'd also avoid attempting lead guidance, and just go for 3 point every-time... this will eliminate any residual chance of intercepting a target at close range as this is inefficient against a crossing target, but at the longest range intercept will make little difference and eliminate problems of incorrect altitude/range setting.

With the forward quarter intercept, chaff is not a problem for the fuse - the target will always be closer than chaff is - unless you have already missed. Set proximity fusing on at launch and be done with it.

The only difficulty is guessing the appropriate range for the altitude/speed/course that the SR71 has, and which can be modified after locking-on or firing. Spreading the 3 missiles over the probable range bracket should allow at least one to approach within fusing range, but actually passing close enough to kill is a harder problem.

Note that the Vietnamese only made 4 engagements against the SR71 in 1967/1968, and one was terminated without missile launch.
Of the 3 remaining engagements, against two separate flights, one missile fused and caused fragmentation damage... depending on whether you count the number of targets engaged, the number of firing systems, or the number of missiles fired, this is between 50%, 33% and 11% hit rate, but a zero kill rate. The only aircraft of this type lost were to operational accidents.