In these checks, the flag they return is relevant to a whole sequence. We should really flag each value according to the flags returned by running the check on ALL relevant sequences that include it, with a defined precedence order between the flags. This is already done for monotonic_decrease_check_cache in #55, but should be done for flatline_check_cache too.
A concern this adds though, is now we need trailing values as well has leading values to QC data with these checks, which stops them being usable in fresh pipelines. I propose we add a variant of these cache checks that pads the data with trailing Nones, so that fresh checks can still Fail data that meets the condition with its predecessors, and flag as DataMissing otherwise, to eventually be overwritten by a periodic pipeline that does have the trailing points. This needs to be done for both monotonic_decrease_check_cache and flatline_check_cache.
In these checks, the flag they return is relevant to a whole sequence. We should really flag each value according to the flags returned by running the check on ALL relevant sequences that include it, with a defined precedence order between the flags. This is already done for
monotonic_decrease_check_cachein #55, but should be done forflatline_check_cachetoo.A concern this adds though, is now we need trailing values as well has leading values to QC data with these checks, which stops them being usable in fresh pipelines. I propose we add a variant of these cache checks that pads the data with trailing
Nones, so that fresh checks can stillFaildata that meets the condition with its predecessors, and flag asDataMissingotherwise, to eventually be overwritten by a periodic pipeline that does have the trailing points. This needs to be done for bothmonotonic_decrease_check_cacheandflatline_check_cache.