为容器映像实现 SnapStart 挂钩

运行快照前挂钩并触发快照流程：在函数初始化代码的最后一步，按需执行快照前挂钩，触发快照流程。仅当 SnapStart 已启用时才执行上述步骤，可通过校验 AWS_LAMBDA_INITIALIZATION_TYPE 境变量的值是否等于 snap-start 进行判断。执行已注册的快照前挂钩，再调用 GET /runtime/restore/next 触发快照流程。若快照前挂钩抛出异常或返回错误，运行时会将该错误上报至 /runtime/init/error 端点。请参见下方伪代码示例：

# After all initialization code has finished:

    READ initialization_type FROM environment variable "AWS_LAMBDA_INITIALIZATION_TYPE"

    IF initialization_type IS "snap-start" THEN

        TRY
            EXECUTE registered before-snapshot hooks
        ON ERROR
            POST error to /runtime/init/error
                SET header  Lambda-Runtime-Function-Error-Type  TO  <Category>.<Reason>
                SET body    TO  { errorMessage, errorType, stackTrace }
            EXIT process with non-zero code

        # Signal readiness for snapshot
        SEND GET request to /runtime/restore/next
        # The request blocks until Lambda restores the execution environment from the snapshot, then returns HTTP 200.

    END IF

运行恢复后挂钩，随后进入调用循环。当 GET /runtime/restore/next 返回 200 时，运行时必须先执行全部已注册的恢复后挂钩，之后才能进入调用循环。若恢复后挂钩执行失败，请向 /runtime/restore/error上报错误。恢复后挂钩执行完毕后，调用 GET /runtime/invocation/next 进入标准调用循环。自此之后，函数行为与未启用 SnapStart 的函数完全一致。请参见下方伪代码示例：

# After the snapshot has been restored
# (i.e., GET /runtime/restore/next has returned HTTP 200):

    TRY
        EXECUTE registered after-restore hooks
    ON ERROR
        POST error to /runtime/restore/error
            SET header  Lambda-Runtime-Function-Error-Type  TO  <Category>.<Reason>
            SET body    TO  { errorMessage, errorType, stackTrace }

# Proceed to the invoke loop