Return a random vector of length n, filled with random elements of type a generated by the mersenne-twister.

E.g. to compute the sum of 100 million random Double values in a vector:

 import qualified Data.VectorUnboxed as U
 import System.Random.Mersenne.Pure64
 import qualified Data.Vector.Random.Mersenne as G
 
 main = do
     g <- newPureMT
     let a = G.random g 10000000 :: U.Vector Double
     print (U.sum a)

The generator will fuse under stream fusion, so e.g. sum . random g will allocate no intermediate array.

Class of types that we have efficient generators for.