holt winters algorithm db: account for replica evaluations not in db

[igorshevach]

Is your feature request related to a problem? Please describe.
today evaluations which are made while (dbModel.IntervalsPassed < model.PerInterval) are not represented in db (skipped).
since holt-winters is a runtime-expensive operation (especially when seasonality period is as high as a week), model.PerInterval can be quite high (e.g. model.tickInterval = 30
sec,model.PerInterval = >15 min / model.tickInterval ). This leads to information missing from the model.

Describe the solution you'd like
while it is important to only run algorithm at sufficiently large intervals of time, the missing evaluations should be represented in db statistically.

Describe alternatives you've considered
extend db model to keep evaluations in between. once a run iteration kicks in (isRunInterval && isRunType == true)
apply some kind of filter over amassed evaluations, update db with the result and dismiss the evaluations.
Thus, only limited number of evaluations (sec,model.PerInterval-1) will be added to db in addition to model.storedSeasons
types of filter may include:

• mean
• max
• median
• random walk (actually, this is what done today)
• run a script
• etc

Additional context
Add any other context or screenshots about the feature request here.