Build `BalancerLib` for spark

[jubeira]

See https://github.com/sparkdotfi/spark-alm-controller/blob/dev/src/MainnetController.sol and e.g. https://github.com/sparkdotfi/spark-alm-controller/blob/dev/src/libraries/CurveLib.sol.

The ALM proxy holds the funds and can batch arbitrary calls as seen in CurveLib.

Our equivalent would be interacting with AggregatorRouter, settling debts first and then calling swap / add (or approve BPT and remove).

We have:

• simple hop swaps
• multi hop swaps (this category includes wrapping / unwrapping to trade through boosted pools using underlying tokens for example)
• simple add / remove
• boosted add / remove (i.e. add / remove to a boosted pool with underlying liquidity)

We can start working on simple and multi hop swaps using the current aggregator router (it will be refactored with #1424 and #1425, but the interface should be the same).

[jubeira]

Narrowing the scope here.

We want to have the following functionality:

• swaps: single hops with wraps / unwraps (flags)
• Adds and removes for pools and boosted pools
• spark-specific rate limiting (same as curve)

Interface-wise, we can keep it close to what we need for the router. E.g.

    function swapSingleTokenExactIn(
        // Might need router / vault addresses here because it's a library
        address pool,
        IERC20 tokenIn,
        IERC20 tokenOut,
        bool wrapTokenIn,
        bool wrapTokenOut,
        uint256 exactAmountIn,
        uint256 minAmountOut,
        uint256 deadline, // Double check whether this is needed in this case
        bytes calldata userData
    ) external returns (uint256 amountOut);

And then use the aggregator batch router if needed to do a 1-hop swap with wraps and unwraps under the hood and settle.

We already can do this for swaps and regular pools; we need to make an aggregator version of the composite liquidity router for boosted pool adds / removes.