The only way for FCP to stitch separate shots is if it is shot in a format that is supported by the Log and Transfer function, and if the camera is designed to merge the clips into one shot via the metadata. For example, P2 breaks up the footage into 4GB chunks, due to the P2 card format's 4GB file size limit. AVCHD will do the same thing, as will XDCAM. But the metadata recorded by the camera informs the NLE that they are essentially one clip, so when you convert via Log and Transfer, FCP (And Avid, and Adobe) will stitch them together.
But this only works with supported formats. H.264 from DSLRs...not supported...by any NLE. Even though you can L&T Canon 5D Mark 2 footage (Mark 3 is unsupported, as it is a newer camera, and FCP was discontinued before it came out), it still keeps the media separate. Sorry...that's the drawback to shooting DSLRs. The ones that shoot AVCHD...will be stitched, because they have advanced metadata support.
FCP-X will multiclip these separate files, however. Because it doesn't have issue with broken clips. It's the best multicam function of any NLE.