use dates.jl for 10x+ benchmark for mdy, dmy, ymd and _hms versions

[drizk1]

improve initial parsing to better leverage Dates.jl and bring benchmarks nearly inline w dates.jl, without sacrificing flexibility for

• mdy_hms, mdy
• ymd_hms, ymd
• dmy_hms, dmy

for ex

@benchmark @mutate(dates, test = Date(dt2, "dd-mm-yyyy"))
BenchmarkTools.Trial: 1398 samples with 1 evaluation per sample.
 Range (min … max):  2.927 ms … 258.430 ms  ┊ GC (min … max): 0.00% … 98.37%
 Time  (median):     3.219 ms               ┊ GC (median):    0.00%
 Time  (mean ± σ):   3.573 ms ±   6.860 ms  ┊ GC (mean ± σ):  8.23% ±  9.48%

   █▆▄▂                                                        
  █████▆▅█▅▆▃▂▂▂▂▂▂▂▂▂▂▁▂▁▁▂▁▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▂▂▂▂▂▂▂▂ ▃
  2.93 ms         Histogram: frequency by time         7.6 ms <

 Memory estimate: 1.42 MiB, allocs estimate: 25145.

@benchmark @mutate(dates, test = dmy(dt2))
BenchmarkTools.Trial: 1287 samples with 1 evaluation per sample.
 Range (min … max):  3.163 ms … 250.221 ms  ┊ GC (min … max): 0.00% … 98.13%
 Time  (median):     3.365 ms               ┊ GC (median):    0.00%
 Time  (mean ± σ):   3.882 ms ±   6.947 ms  ┊ GC (mean ± σ):  7.76% ±  9.09%

  ▅█▆▅▅▃▂▂▃▂▁                                                  
  ███████████▇▇▆▆▇▅▇▆▆▆▅▆▄▁▄▅▅▁▄▄▄▁▁▁▁▄▁▄▄▁▄▄▄▁▄▄▆▅▅▇▅▄▆▄▄▄▁▅ █
  3.16 ms      Histogram: log(frequency) by time      8.15 ms <

 Memory estimate: 1.44 MiB, allocs estimate: 26145.