Portfolio weights and clustering

Toolset perf_weights: 4 tools.

Tool	What it does	Notes
`perf_weights`	Portfolio weights from the assets’ returns: inverse volatility, mean-variance (the highest return per unit of risk within bounds, on a shrunk or sample covariance) or equal risk contribution (risk parity, with optional risk budgets); optionally capped at a maximum weight with the excess redistributed. Answers the weights and the resulting portfolio’s annualised return and volatility over the same sample.
`perf_limit_weights`	Caps every weight at a limit and redistributes the excess to the other assets in proportion to their weights, repeating until none is above it.	No price input.
`perf_random_weights`	Draws random portfolio weights within per-asset bounds that sum to a total, to benchmark a strategy’s weights against.	No price input. Reproducible with `seed`.
`perf_clusters`	Groups assets by the correlation of their returns, with k-means on a two-dimensional map of correlation distances or with the fast threshold clustering algorithm, and answers each cluster’s members with their average correlation inside it, and the correlation matrix.	Reproducible with `seed`.

Inputs, limits and outputs are described in Fincept Performance. Full schemas: fincept_describe_tool.