Microsecond time noise in some AstroEpochs conversions #1
Description
Some transformations are only accurate to microseconds. It appears there are a few issues to address. One is a possible bug in reference epoch used in AstroEpochs for some scales in shift. Another is that shifts for some times are treated as offsets from an epoch far in the past, in seconds. This truncates time precision. We need to always keep the jd2 portion of time (whether handling in offsets in seconds or anywhere else) to less than one day. If correct about the second issue (I am not certain yet), this would require a modification to offset_tdb2tcb.