FEATURE: Budgetize Input amplification

[danglotb]

Characteristics

• Issue Type: feature
• Severity: feature, major

Description

Create a budget for the Input Amplification. Among the list of amplifiers, each of them gains a given amount of budget, e.g. a number of methods to produce. Then, we limit the input amplification according to this budget.

This aims at keeping diversity without wasting resources. In fact, each amplifier will produce the same amount of test methods that allows exploring each different path (path in a sense of exploration path, in the exploration space).

• compute a budget
• predict the consumption of amplifiers
• stop the input amplification according to the budget
• differentiate between random than hard-coded operators, because it is likely that we want to prioritize hard-coded operators rather than random

a side feature is:

• unmark all amplified elements in case of all them have been amplified. However, we should only use random operators to keep doing the exploration.

Help is welcome

[danglotb]

The budget implemented in #477 does not answer to this issue but implement another way.

For now, the SimpleBudgetizer selects a fair number for each Amplifier and each test methods to be amplified. The time consumption seems to be ok for now.