Here are some more technical answers...
The presence of the DV cartridge has nothing to do with the screen colors. Each "layer" of base-case video feeds from its own 512 KB memory area, which makes it impossible to use cartridge memory for base-case video.
And it wouldn't be helpful either; the "normal" screen resolution is one of only a few fixed values (384x280 or 360x280 for PAL, 384x240 or 360x240 for NTSC). You can double this resolution horizontally by using a 4 bit CLUT mode and vertically by using interlacing, but no other variations are possible. The base-case video hardware fetches at most one byte per plane per normal resolution pixel and this cannot be changed. Doubling the default PAL resolution both horizontally and vertically gives the 768x560 resolution you mention, but for NTSC the maximum is only 768x480.
Only a few players can do both NTSC and PAL, the 180 and 60x models being among them. The supported mode is usually determined in hardware; the same system ROMs often support both modes (with CD-i Emulator I found that in some cases a "Philips" player will suddenly change to a "Magnavox" player when you switch it from PAL to NTSC).
The MPEG video "layer" can only display MPEG images (either still pictures or moving video); it is not directly accessible in a documented way.
Most games use the 7-bit CLUT modes in both planes, giving them a total of 256 colors to play with. Sometimes DYUV is used for a special effect or small "movie"; the 16 million colors are delta-coded which makes it impossible to have arbitrary changes in color. RGB555 is hardly ever used; it needs both planes to get the required 15 bit/pixel bandwith from memory (the 16th bit is for transparency control).
If you want more technical information about the base-case video subsystem, I refer you to the Motorola documentation for the Motorola MCD212 Video Decoder and System Controller chip, available in the
CD-i Technical Documentation / System section of the ICDIA website. Most newer CD-i players use this chip and it almost exactly implements the Green Book specifications.
Without sprites you can still do movement; PC video hardware doesn't have sprite support either! You just need to erase and redraw a "sprite" image over a background, generally for each video frame. This is just replicating the sprite overlay function in software...
As I stated, the 4 MB for the 180 and 60x players is
additional RAM; together with base-case (1 MB) and cartridge (1.5 MB) RAM these players can have up to 6.5 MB total.
CD-Bridge is a CD-XA profile that requires a CD-i application on the disc; VideoCD and PhotoCD are both CD-Bridge applications. This was intended to produce PC compatible CDs, but still requires PC software appropriate for the disc format (this was not normally included on the CDs, although it could be).
You can use DYUV (or CLUT, for that matter) to play "small" videos: it's just a matter of redrawing the appropriate images fast enough. The bottleneck is usually the CD bandwidth, which is at most 170 KB per second as the CD drive runs at single-speed only. Don't know about Voyeur, but Burn:Cycle uses some very clever run-length compression for its videos: they managed to make an artistic asset out of the bandwidth limitations. The video hardware supports run-length decompression for the CLUT3 and CLUT7 modes (8 and 128 colors, respectively; the highest bit is used to control the runs).
I have no answers for your other questions. Perhaps someone else here does?