Refactor skills.json

[austinworks]

As a datastore we are using for testing purposes, maybe we can make this data more human readable. This might also improve the LLM's ability to accurately complete tasks.
Current version is 6000+ lines with much repeated data.

continuous-alignment-testing/examples/team_recommender/tests/fixtures/skills.json

Lines 6083 to 6098 in db2b44e

	{
	"developerId": 16,
	"skillId": 28,
	"skillLevel": 3,
	"comment": "on a government mobile engagement in 2023",
	"developer": {
	"id": 16,
	"email": "morganbrown@artium.ai",
	"name": "Morgan Brown",
	"archivedDate": null
	},
	"availability": {
	"startDate": "2025-02-06T00:00:00Z",
	"endDate": "2025-06-01T00:00:00Z"
	}
	},

Maybe we should just be using SQLite or something relational for representing relational data.