John,
The term preventive maintenance (PM) is a legacy from the 1950's Manufacturing Industry, particularly the automobile plants. The concept was to 'prevent' failures through a hard-time task, replacing a susceptible component well before it failed and stopped a whole production line. This alone was considered pro-active.
Then PdM came along, allowing us to predict time-to-fail for these susceptible components. That allowed us to plan and schedule the replacement, addiing to our list of pro-active tasks.
In the 70's and 80's, RCM influenced our thinking. We could allow items with a low failure consequence to run-to-failure. Such breakdowns are anticipated, because we have decided to let the items fail, so they can be planned, i.e. work steps, resources, spares, logistics etc. identified. But we do not know when exactly they will fail, and hence they cannot be scheduled. In such cases, the planning can be pro-active, but the scheduling is always reactive. In addition, there are breakdowns that we did not anticipate, so there is no plan or schedule, thus reactive on both counts. Since in many Companies, for both types of breakdowns, no plan or schedule exists, your position that any activity not in the weekly schedule is reactive can be supported. For those Companies that do do plan run-to-failure tasks, most of the benefits of pro-active work are there, but there is some loss due to rescheduling the weekly workload. In practice, it is hard to make this distinction from a CMMS, so your concept of treating all unscheduled activities as reactive is justified.
The problem has always been the term PM, meaning time-based maintenance. It is true that PM is pro-active, but so is PdM, and in theory, some run-to-failures too, because that is what WE decided to do, not what the machine decided to do (refer my earlier post).
All work that we decided to do in advance and planned are pro-active, the rest is reactive. Ron Moore in his book "Making Common Sense..." provides data to suggest that the best performers limit reactive work to 10%. That is a good target, IMHO.