relocate parser look ahead loops

[jerch]

Part of #1823.

This PR relocates the read ahead loops into the proper switch cases. It further reduces code complexity (less branching) with a slight speedup for the optimized actions.

[jerch]

Speed comparison (tested with this script):

• before:

   Context "examples/parser_throughput.js"
      Context "Parser performance - 50MB data"
         Context "print - a"
            Case "throughput" : 5 runs - average throughput: 432.44 MB/s
         Context "execute - \n"
            Case "throughput" : 5 runs - average throughput: 95.12 MB/s
         Context "escape - ESC E"
            Case "throughput" : 5 runs - average throughput: 53.92 MB/s
         Context "escape with collect - ESC % G"
            Case "throughput" : 5 runs - average throughput: 22.37 MB/s
         Context "simple csi - CSI A"
            Case "throughput" : 5 runs - average throughput: 84.30 MB/s
         Context "csi with collect - CSI ? p"
            Case "throughput" : 5 runs - average throughput: 55.03 MB/s
         Context "csi with params - CSI 1;2 m"
            Case "throughput" : 5 runs - average throughput: 104.91 MB/s
         Context "osc (small payload) - OSC 0;hi ST"
            Case "throughput" : 5 runs - average throughput: 50.15 MB/s
         Context "osc (big payload) - OSC 0;<text> ST"
            Case "throughput" : 5 runs - average throughput: 48.93 MB/s
         Context "DCS (small payload)"
            Case "throughput" : 5 runs - average throughput: 49.25 MB/s
         Context "DCS (big payload)"
            Case "throughput" : 5 runs - average throughput: 94.54 MB/s

• after:

   Context "examples/parser_throughput.js"
      Context "Parser performance - 50MB data"
         Context "print - a"
            Case "throughput" : 5 runs - average throughput: 632.97 MB/s
         Context "execute - \n"
            Case "throughput" : 5 runs - average throughput: 89.43 MB/s
         Context "escape - ESC E"
            Case "throughput" : 5 runs - average throughput: 58.36 MB/s
         Context "escape with collect - ESC % G"
            Case "throughput" : 5 runs - average throughput: 22.37 MB/s
         Context "simple csi - CSI A"
            Case "throughput" : 5 runs - average throughput: 89.36 MB/s
         Context "csi with collect - CSI ? p"
            Case "throughput" : 5 runs - average throughput: 94.44 MB/s
         Context "csi with params - CSI 1;2 m"
            Case "throughput" : 5 runs - average throughput: 124.12 MB/s
         Context "osc (small payload) - OSC 0;hi ST"
            Case "throughput" : 5 runs - average throughput: 50.08 MB/s
         Context "osc (big payload) - OSC 0;<text> ST"
            Case "throughput" : 5 runs - average throughput: 48.31 MB/s
         Context "DCS (small payload)"
            Case "throughput" : 5 runs - average throughput: 60.97 MB/s
         Context "DCS (big payload)"
            Case "throughput" : 5 runs - average throughput: 256.35 MB/s

Note there are still two quite slow actions - "OSC" and "ESC with collect". Both do intermediate string conversion + concatenation, maybe we can get rid of that with later PRs.

[Tyriar]

LGTM after the comment is removed