accumulate total tokens spent referenced with prompt names

[github-actions[bot]]

@h1alexbel the issue body is empty, please provide more details for this problem.

[h1alexbel]

let's accumulate the total token spent of the LLM we use. I suggest to stick to this formatting:

total token spent: 7

analysis: 1
validate: 1
validate.json: 1
top: 1
candidate.json: 1
suggestions: 1
suggestions.json: 1

we even can render more attractive graph or table with this information