feat(query): add Boost parameter to all query types

[bevzzz]

This PR adds a Boost parameter to all query types. The API mirrors the reference implementation in the Python client.

var boost = Boost.filter(Filter.and(...), b -> b.weight(.1f));
var boost = Boost.timeDecay("last_seen", "14d", b -> b.origin("2024-01-01T00:00:00Z"));
var boost = Boost.numericDecay("size", 36f, .55f, b -> b.curve(Boost.Curve.LINEAR));
var boost = Boost.numericProperty("size", b -> b.modifier(Boost.Modifier.LOG1P).depth(2f));

c.query.nearVector(v, opt -> opt.boost(boost));

N.B.: a big chunk of the diff is the re-generated WeaviateProtoSearchGet (~6.2K), which can be safely ignored.

Closes #576

[orca-security-eu]

Orca Security Scan Summary

Status	Check	Issues by priority
Passed	Infrastructure as Code	0   0   0   0	View in Orca
Passed	SAST	0   0   0   0	View in Orca
Passed	Secrets	0   0   0   0	View in Orca
Passed	Vulnerabilities	0   0   0   0	View in Orca

[trengrj]

LGTM, just one comment the python client checks that depth is not set on sub boosts. Would be good to also validate this in the java client (couldn't find if this exists).

        if len(boosts) == 0:
            raise WeaviateInvalidInputError("Boost.blend() requires at least one boost.")
        for r in boosts:
            if r.depth is not None:
                raise WeaviateInvalidInputError(
                    "Cannot set `depth` on sub-boosts passed to `blend()`. Use the top-level `depth` parameter instead."
                )

[bevzzz]

@trengrj added that validation 👍

As a general rule, I try to let the server validate the request and not duplicate that logic on the client as well; this is a good example of a situation in which the server wouldn't even known depth was set on a null sub-boost, so this can only be validated client-side.