Any ideas for files where "rep" coding could make more sense?

Code:
enwik8, 100M bytes, compressed file length = 24560059 
 - 13 bytes lzma header - 1 byte rc pad = 24560045

     Submodel  log2_entropy   rc_output
------------------------------------------
    c_IsRepG2:     1879.37      1879.37
    c_IsRepG1:     5340.18      5340.18
    c_IsRepG0:     9346.86      9346.91
 c_IsRep0Long:    14287.23     14287.22
      c_IsRep:   149490.93    149497.97
    c_SpecPos:   356311.28    356311.19
   c_Literal2:   415920.15    415922.79
 c_LenChoice2:   445556.96    445556.61
    c_LenHigh:   589577.69    589580.39
     c_LenLow:   768648.29    768648.04
  c_LenChoice:   809017.17    809016.98
     c_LenMid:  1019553.67   1019553.34
    c_IsMatch:  1117219.48   1117219.17
   c_Literal1:  1543005.86   1543010.79
      c_Align:  3115007.86   3115007.06
    c_PosSlot:  4310701.82   4310706.69
  direct bits:  9889156.75   9889156.78
------------------------------------------
        Total: 24560021.55  24560041.48
Code:
icl.exe, 3128224 bytes, compressed file length = 996740,

     Submodel  log2_entropy   rc_output
----------------------------------------
 c_IsRep0Long:      257.64       257.63
    c_IsRepG2:     1361.74      1361.74
    c_IsRepG1:     3656.54      3656.54
    c_IsRepG0:     5445.74      5445.74
 c_LenChoice2:     7072.96      7072.96
     c_LenMid:    12381.42     12381.42
    c_LenHigh:    13681.88     13682.00
      c_IsRep:    24667.16     24667.10
  c_LenChoice:    26850.71     26850.68
    c_SpecPos:    27668.61     27668.60
   c_Literal2:    57264.92     57264.93
    c_IsMatch:    72904.86     72904.86
     c_LenLow:    89370.24     89370.24
      c_Align:    96537.44     96537.42
    c_PosSlot:   148082.87    148083.39
  direct bits:   153976.38    153976.38
   c_Literal1:   255540.85    255540.80
----------------------------------------
        Total:   996721.96    996722.44