Unless you're running software which specifically wants a math coprocessor and/or you plan on running alternative operating systems (A/UX, NetBSD, etc) 68LC040 vs "full" 68040 hardly matters at all. If you bought the thing just to fiddle with old Mac games on I wouldn't obsess over it. For integer tasks (which is most game and productivity software) the two CPUs will execute code at exactly the same speed.
(The aforementioned "alternative operating systems" require a "full" '040 because there is a hardware flaw with most LC CPUs that prevents FPU emulation software which depends on inline trapping of the unsupported instructions from working entirely. This doesn't affect most MacOS programs because the math functions which would use FPU are routed through a higher-level library in the first place. There probably is a small category of ill-behaved MacOS programs that contain inline FPU instructions but I can't name any offhand.)
(The aforementioned "alternative operating systems" require a "full" '040 because there is a hardware flaw with most LC CPUs that prevents FPU emulation software which depends on inline trapping of the unsupported instructions from working entirely. This doesn't affect most MacOS programs because the math functions which would use FPU are routed through a higher-level library in the first place. There probably is a small category of ill-behaved MacOS programs that contain inline FPU instructions but I can't name any offhand.)