tracks guy

That can't be very precise. In that diagram, for instance NTSC colors 1E and EE are exactly the same, RGB = FEFA40. The diagram is called "NTSC (128 unique colors)" :-)

How close to linear is the luminance output by a tv with respect to the luminance bits field of color registers in the 2600? That is, can I expect the total light from one pixel of e.g. color $2E and one pixel of color $26 to consistently closely match the total light from two pixels of color $2A? The reason I ask is I am working on anti-aliasing graphics, where part of an object sometimes are represented by some light in multiple pixels, and sometimes just light in one pixel. If I can see that it doesn't flicker too much on my tv and in stella, does that mean I can assume it will work well for all monitors?

If you don't mind a few pushes to the stack, this is 5 bytes and 63 cycles: lda #0x0A ; 0A is the ASL instruction pha bne *-2

Yeah... here blue and yellow is playfield, green and orange is P1, red is P0.

Here is a riddle for you: How do you make 6 consecutive pixels have 6 different colors? Anyway... lots of colors, a 40 pixels wide great role model, and some mystery. Who could ask for more?