general graphx optimizations

[ZERICO2005]

smaller optimizations across graphx:

• optimized SMC code for gfx_floodfill
• removed redundant comparison check in one of the while loops for floodfill
• optimized off screen sprite clipping by returning directly to the caller if the sprite is off-screen. This makes all paths a minimum of 1F + 2 faster.
• switched _indcallHL to ti._indcall to save a relocation
• saved 1 byte in gfx_SetCharData
• optimized gfx_TilePtr via djnz hax
• ld iy, 0 \ ld iyl, a \ ld iyh, SMC can be changed to inc.s iy \ ld iyl, a \ ld iyh, SMC
• the condition for .positive_dy was checked twice instead of once, and the comparison only needed to use 8 bits instead of 16/24 bits, so I was able to optimize that for gfx_Line_NoClip

11795 - 11745 = 50 bytes saved