In thinking about it more, I realized the heat sinking would be an issue. But I found a solution. Rather than buying new Luxeons, why not just use two 1157 sockets in the PCB board and use the functional Luxeons I already have! It would simplify the design and make construction easier. They are already have built in heatsinks and 1157 sockets are cheap. Since they are largely directional anyways, I doubt the reflectors are doing much, so those could be gotten rid of and just angle the 1157 sockets like 5-10 degrees of the front/back axis to get a slight spread.
For the details on the the directional arrow circuits, I got to thinking you could actually add the motion pretty easily. Divide the two/three/four rows of leds up and put an SCR on each line. The SCR would work that once tiggered, it would keep the "chevron" illuminated until the power was cut. If the SCR's could be progressively timed such that the innermost (#1) turned on after 100ms, (#2) at 200ms delay, (#3) at 300 ms delay, and optional #4 at 400 ms delay, you could get a really neat effect.
every time the turn signal relay turned on, you would get a motion effect on the directional arrow of LED's as it progressively illuminated the chevron.
however, I don't know how do to timed triggers for SCRs. Any EE's out there wanna help?? Lou? I think a RC circuit connected to a UJT to signal each SCR could do it.
Acutally, you could do it with only one trigger circuit serving either side. Have the trigger ciruit powered by EITHER the left or right supply, while the SCRs are tied only to their respective left/right supply. The 100ms trigger would send a signal to both the left and right #1 chevron SCR, but only the appropriate only would have a supply voltage to turn on. Additional on hazzards you'd get a nifty outwards spreading effect on both sides.
100ms seemed right, but 150ms spacing might be more appropriate.
I'm definitely going to make a prototype if anyone can supply the trigger circuitry!!