A mapping worked only with exact matches which means you couldn't directly apply the lookup like shown. Are there any range-areas needed it requires to extend the mapping-table for them respectively to resolve the ranges to dedicated values.
A quite common approach for such scenarios is the use of internal while-loops, like:
m: mapping load Key & '|' date(Start + iterno() - 1), Return
from Source while Start + iterno() - 1 <= End;
If it's possible with your data to resolve the Order & Date in this way and/or applying other measurements to add extra information like the min/max values and/or a count and/or any accumulation ... ? Further useful and important is that mappings could be nested - in the table as well as within the call. This means multiple lookup-values could be loaded for a single query and also multiple queries against a single lookup-value.
From a logically point of view the above loop-example is already such a nesting because each Key is combined with n dates to ensure respectively to increase the likely that the query would find a match. The reverse is possible within the query with something like:
applymap('m', Key & '|' today(), applymap('m', Key & '|' date(today()-1), ...))
By each kind of nested approaches the order of the execution is important and could be targeted used to get the wanted results.
