ACD_Evidence now tests each key comparison with a Diebold-Mariano test
using Newey-West variance over an hour of snapshots, because the forecast
targets overlap. Two new rungs repeat E and H with the activity ratio
alone, showing that the plotted line carries the gain by itself.
Also sync AcdDiurnal.mqh with the spelling used in the article listing,
reword the feed-check comment in ACD_Calibrate, and update the README
with the new findings.
Companion code for the MQL5 article on autoregressive conditional
durations: a five-file library, the calibration and evidence scripts,
and the live activity indicator.
