Racket*:refactor the code

[Antigen-1]

• In order to reduce the number of intermediate containers, Racket ,Racket_compiled and Racket_parallel all implement a priority queue without concurrency involved. We simply don't need a Racket_concurrency version anymore so I remove it in this commit.
• Racket and Racket_compiled share codes with Racket_parallel.
• I replace string->number with two simpler functions, digits? and bytes->integer, to avoid checking. So Racket_parallel is faster than before, but it's still slower than Racket_compiled. It is very expensive to create a place, so currently the number of places is set to 2.

Here are results of run_benchmark:

Running Racket_parallel...Ok(BenchmarkResult { topavg: 590335206, avg: 602667206, min: 575964646, max: 627359418, mdev: 14567870, count: 20 })
                 Python3: top:  233.65, min:  223.55, avg:  250.96, max:  296.52, mdev:   20.26, cnt:  20
         Racket_compiled: top:  411.08, min:  409.84, avg:  417.80, max:  488.91, mdev:   17.54, cnt:  20
              ChezScheme: top:  461.49, min:  451.06, avg:  474.71, max:  511.36, mdev:   16.25, cnt:  20
         Racket_parallel: top:  590.34, min:  575.96, avg:  602.67, max:  627.36, mdev:   14.57, cnt:  20

[Antigen-1]

Racket_parallel is faster than ChezScheme now.

                 Python3: top:  208.80, min:  208.09, avg:  223.26, max:  242.51, mdev:   15.88, cnt:  20
         Racket_compiled: top:  433.69, min:  432.43, avg:  438.53, max:  476.33, mdev:   10.67, cnt:  20
         Racket_parallel: top:  559.78, min:  541.55, avg:  579.60, max:  630.60, mdev:   25.17, cnt:  20
              ChezScheme: top:  576.07, min:  430.11, avg:  647.61, max:  789.79, mdev:  106.02, cnt:  20