Try replacing the (wiimote) version of groovymame supplied with the iso in /usr/local/bin/ with this one:
https://54c0ab1f0b10beedc11517491db5e9770a1c66c6.googledrive.com/host/0B5iMjDor3P__aEFpcVNkVW5jbEE/v0.171_015m/groovymame64_0171.015m_linux.tar.bz2I can confirm Arkanoid is working normally and Warlords is working normally with the sensitivity back at 50 (although I need it at 100 for gameplay purposes).
From what I can tell, setting the sensitivity to 50 (in Warlords, at least) makes the sprite move only half the range of the pointer. That is to say if the sprite and the (invisible) mouse pointer both start at full left, then when you move the pointer full right it gets stuck at the boundary of the screen, as expected, but the sprite has only moved half way and any extra attempted movement isn't added to the sprite's position. If you then use a digital method to move the sprite a bit furter right, say to three quarters, then you can move the pointer all the way back but when it sticks on the left hand side the sprite will only move half of its travel and stop one quarter right, allowing movement from one quarter to three quarters across the screen. It seems like the sprite position is derived from the pointer position and not the input speed of the device.
I can set up several pointers with xinput but the no wiinote version bunches them all together and they act as one. So using one pointer to move the sprite to the right a bit then using another pointer to move it further transitions smoothly. The sprite moves, stops when you swap to the next mouse then moves on further. There doesn't seem to be a block when the pointer hits the edges of the screen, although the pointer will stop, the extra movement beyond the edge is still added to the sprite's position, allowing full movement of the sprite no matter where the pointer starts off. I think the movement of the sprite is taken directly from the movements of the mouse rather than the position of the pointer (which seems like a better way to me) but it isn't differentiating between the mouses. Windows has only one pointer but can differentiate between inputs and the sprites can be moved independently and simultaneously.
The wiimote version's behavior is totally different. It seems to detect all the pointers but can only handle one at a time and still bunches them all together. Say pointer 1 starts on the left and pointer 2 is in the middle of the screen. Assume the sensitivity is set for full screen movement. If I use pointer 1 to move the sprite it is fine but it I switch to pointer 2 the sprite will teleport to the middle (pointer 2's initial position) and continue moving normally. If I switch back to pointer 1 the sprite will teleport back to where it was when pointer 1 left it. Again, it gives me the impression that the sprite's position is derived from the pointer position.
The actual pointer when activated in mame can be seen to teleport about in this manner as well, wheras in X separate pointers are shown and can be moved independently and simultaneously without interfering with another.