Increasing of L doesn't help on linearity, you just move operating point to less steeper point of the square law characteristic. However degeneration uses the high non-linear gain of the diff.pair to introduce a very linear gain by negative feedback. Negative feedback is necessary to reduce distortion, and degeneration is a very simple way to implement it.
And source degeneration of diff. pair doesn't increase Vgs like increasing of the length, so it cannot ruin the input common mode voltage range (just an other reason, no connection with linearity).