It probably depends on the domain. For telescope people, a “signal” is anything other than noise and systematic error in your measurements.

I’d speculate that this terminology came from telephony or radio communications - even the term “noise” is a giveaway. But it has migrated into astronomical language in a slightly different way.