The captures are good. We get
{267}e1e1e1e1e1e1e1e1e1e1e1e1fccd3532d532d55552d52d532accccb535555552d200
{218}e1e1e1e1e1e1fccd3532d532d55552d52d532accccb4d555554d2c00
{219}e1e1e1e1e1e1fccd3532d532d55552d52d532accccb4b555554b4a00
{219}e1e1e1e1e1e1fccd3532d532d55552d52d532accccb355555534d200
{219}e1e1e1e1e1e1fccd3532d532d55552d52d532accccd355555534b200
shifted by one bit this is
70f0f0f0f0f0f0f0f0f0f0f0fe669a996a996aaaa96a96a99566665a9aaaaaa9690
70f0f0f0f0f0fe669a996a996aaaa96a96a99566665a6aaaaaa6960
70f0f0f0f0f0fe669a996a996aaaa96a96a99566665a5aaaaaa5a50
70f0f0f0f0f0fe669a996a996aaaa96a96a995666659aaaaaa9a690
70f0f0f0f0f0fe669a996a996aaaa96a96a995666669aaaaaa9a590
The decoder looks for a fixed sync of
f0f0f0f0f0f0f0ffBut the best we get here is 0f0f0f0f0f0fe
So this is just an instance where the first databit is zero instead of the assumed one. Also the sync search is too harsh.
Please open an issue at rtl_433 Github, with at least one broken sample file so we can track and fix that. Thanks!