chore: Align log format with upstream implementation

[hassila]

Should kill the dependence on TextTable as part of that (sorry, mea culpa) - will also make the package more palatable to depend on as there will be fewer dependencies (should probably remove SPI manifest checker bundle too).

Also, we seem to have a truncating error in the value output?

If looking at the upstream sample file

    Value   Percentile   TotalCount 1/(1-Percentile)

    0.016     0.000000            1         1.00
    4.455     0.100000       109166         1.11
   10.367     0.200000       218441         1.25
   14.383     0.300000       327522         1.43
   19.471     0.400000       436590         1.67
   30.447     0.500000       545740         2.00
   39.519     0.550000       600381         2.22
   52.191     0.600000       654888         2.50
   68.223     0.650000       709558         2.86
   87.807     0.700000       763927         3.33

The value seems to be interpolated for the percentiles? We always get '000'.

        Value     Percentile TotalCount 1/(1-Percentile)
------------ -------------- ---------- ----------------
      11.000 0.008333333333          1             1.01
      56.000 0.150000000000         18             1.18
     109.000 0.200000000000         24             1.25
     213.000 0.300000000000         36             1.43
     305.000 0.400000000000         48             1.67
     441.000 0.500000000000         60             2.00
     471.000 0.550000000000         66             2.22
     528.000 0.600000000000         72             2.50
     548.000 0.650000000000         78             2.86
     601.000 0.700000000000         84             3.33
     700.000 0.750000000000         90             4.00

[dimlio]

Also, we seem to have a truncating error in the value output?

Probably they used scaling factor when printing distribution?

Anyway I made output identical to one from Java implementation.
(Tested by comparing percentile output for some random data).