Long range AA missile intercept geometry

Long range anti air missiles occasionally follow very strange and unrealistic intercept paths.
Looks almost as if they were flying to a precalculated intercept point and then turn towards the target, which could be normal for long range AA but this behaviour is not consistent (usually the missiles are tracking their targets all the way, see Gammon example) and the intercept points also seem wrong, like too far away, often beyond the target.

The targets are often loitering, the problem could be linked to the loitering-at-cruise-speed issue.

I could not yet recreate this with a test scenario (tried but missiles work ok in the small test scenarios) but here are 3 examples with battleset scenarios.

Phoenixgeom:
Group AIA launches Phonixes at group ZNA, missile group ALM flies strange zig-zag course to ZNA, distinctly changing course two times.

Phoneixgeom2:
Group AIA launches Phonixes at group ZNA,
ZNA loiters on high but with 590kts speed (e.g. it's cruise speed)
when AIA launches ZNA changes it's course to 48 degree from 209 degree,
missile group ALM flies *over* and passes ZNA, then turns back to it.
(The missiles' intercept point looks like it could be the point where ZNA could have been if it had been really cruising.)

Gammongeom2:
Missile group ZLM is attacking AHA, flies to a point where AHA was a minute earlier, then turns to track AHA.
Gammongeom1.hp8 is a savegame from 1 minute earlier, before AHA reached it original patrol destination and started to loiter, however I can not recreate the situation recorded in Gammongeom2.hp8. Still the issue can be consistently observed in Gammongeom2.