feat: add amountMath.find

[katelynsills]

This PR adds amountMath.find.

amountMath.find takes in a leftAmount and a searchAmount. It returns a new amount that represents the parts of leftAmount that "match" the value of searchAmount. If nothing matches, the identity element is returned. For example:

const leftAmount = make(harden(['abc', 'ab', 'b']));
const searchAmount = make(harden(['a']));
find(leftAmount, searchAmount);
// returns equivalent of make(harden(['abc', 'ab']));

and

const leftAmount = make(harden([{ name: 'a', handle: {} }, { name: 'b', handle: {} }]));
const searchAmount = make(harden([{ name: 'b'}]));
find(leftAmount, searchAmount)
// returns equivalent of make(harden([{ name: 'b', handle: {} }]))

This method can be used to find an amount in leftAmount using partial descriptions in searchAmount, such as string prefixes in the case of MathKind.STRING_SET or keys and values with some keys and values omitted in the case of MathKind.SET. If searchAmount.value is an array as in the case of MathKind.STRING_SET and MathKind.SET, every element of the array must have a match or the identity element is returned. Additionally, if searchAmount.value is an array, a separate search is performed for each element in searchAmount.value and the results are returned in an array with any duplicates among the searches removed.

This PR adds a number of tests, but further usages of amountMath.find are left out. I will follow up with a few PRs stacked on this one.

Business value with further usages:

• Enables support for denoms (User Story: Specify want with regexp pattern, including string prefix #1913)
• Enables support for flexible wants in Zoe (Want not exact (e.g., any 3 tickets) #1915)
• Enables searching for and withdrawing an invitation from the wallet without knowing the invitation handle (Withdraw an invitation from the wallet by a search not based on invitationHandle #2120)

Changes to setMathHelpers

It was significantly easier to implement this by reducing the flexibility of setMathHelpers. Instead of allowing arrays of anything comparable, I restricted setMathHelpers to only operate on arrays of records with string keys.

Related #1905

[erights]

Just comments so far. I need to see how this is used before I can make substantive comments.

[erights]

Top level mutable state make this module impure. Is this intentional?

[erights]

Unfortunately, for JavaScript this is probably the best way to do this. In tc39 I had suggested generalizing break <label>; so that you could break to a label in a lexically enclosing function, giving us a non-erroneous non-local exit. It would make JS more expressive and arguably simpler at the same time, since it "simply" removes a restriction that would no longer need to be explained. Ah well.

[erights]

This is frugal. As used by eventualWithdraw you're withdrawing the least amount of rights that satisfies the pattern. This is opposite of the set variants, which leaves me confused about what the abstract meaning of the operation is.

[erights]

This is anti-frugal. As used by eventualWithdraw you're withdrawing the most amount of rights that satisfies the pattern. This is opposite of the nat variant, which leaves me confused about what the abstract meaning of the operation is.

[katelynsills]

Closing, but we can use this for design in the future. It is also linked from the epic