[ainyan03]

Panel_ED2208::_wait_busy() treated BUSY as released only after it stayed high for 200 ms, assuming a short high between the steps of the AUTO sequence; right after a refresh started it required the 200 ms even when BUSY was already high. The wait now ends at the first high (displayBusy() turns false), which also covers boards without a BUSY pin.

Measurements (PaperColor, room temperature)

BUSY and CS edges were recorded with GPIO interrupts (µs timestamps):

• BUSY goes low while the last byte of AUTO is being received (4-5 µs before CS rises) and stays low for the whole sequence, with no short high, for black/white, 6-color and dithered images (14.7 s each).
• First refresh after power-on, alternating the old and new code 5 times each from a cold start: all refreshes started and completed, no brownout resets.
• Without a reset pin (AUTO 0xA5), three consecutive refreshes with the next one sent within about 1 ms of BUSY going high: all accepted and completed, each image confirmed visually.
• display() takes about 1.12 s instead of 1.33 s (the 200 ms is no longer added to the wait for the previous refresh).

Testing

• Local build matrix of the versions used by this repository's CI: all passed. CI on the fork: passed.