Env refactor

[namesty]

Overview

This PR aims to change the way environment variables/configuration is passed to wrappers. This was done with 5 main objectives in mind:

• Change the wasm-boundary functions to follow from a a "pull" model, instead of a "push" model
• Remove ClientEnv and sanitizeEnv concept
• Support environment configuration for Rust wrappers
• Enable/pass environment at the method level
• Prepare implementation for, in the future, overriding environment properties at the method level

Over the course of the development and discussion process, we realized there were additional tasks to implement to be feature complete. So this PR also enables:

• Support for importing external environment types
• Support for conditional environment configs
• Support for environment configuration for subinvoked wrappers
• E2E tests for both subinvocation with env configurations, and Rust wrappers with env configurations.

Developer experience changes

We're introducing the @env directive, which annotates methods that use an environment configuration:

schema.graphql:

type Env {
  str: String
  requiredInt: Int!
}

type Module {
  getEnv(arg: String!): Env @env(required: false)
}

Additionally, now module methods take a second env argument, from where the wrapper developer can access the environment variables:

index.ts:

import { Env, Input_getEnv } from "./wrap";

export function getEnv(input: Input_getEnv, env: Env | null): Env | null {
  return env;
}