Follow-up to #1156. That PR found working cids (01, 08, 12 - the range
was 02-07,10,11 before) for channels previously marked [NO PUBLIC
STREAM], which is valuable, but it also reassigned the cid of every
already-known-working channel (NHK BS, BS日テレ, BS朝日, BS-TBS,
BSテレ東, BSフジ) to its neighbor's old value - a clean cyclic shift
across exactly six channels, which is what a list/list zip misaligned
by one produces, not what channel-by-channel verification produces.
Neither the old nor the new mapping carries any content-level proof
(the streams have no identifying metadata), so there's no way to tell
which is right - only that the shift itself isn't evidence of anything.
This takes only the part that's unambiguous: three cids nobody had
before, given to the three channels that had nothing. bs07 was one of
the "new" ones in #1156, but it's already BS10スターチャンネル's cid,
so WOWOWシネマ gets bs08 instead - the only other free value.
Verified each new cid resists a single flaky read (checked 4x): bs01
and bs12 came back flaky (3/4 ok) but real, bs08 came back alive (4/4).
That's consistent with how naori-test.netgenx.site behaves generally
(see check_channels.py's docstring), not a sign these three are wrong.
Kálmán „KAMI” Szalai
e7e5ca8a62