Phase 3 resource support: multi-token resources via oneTBB range-based limiter construction #853

Scope

Phase 3 enables multi-token resources: a resource that declares both a token_type and a tokens() member. Each returned token represents one concurrent access slot. A resource offering N tokens bounds a dependent node to at most N concurrent invocations, subject to its configured operator concurrency.

Phase 1 (#851) recognizes this form but rejects it at compile time. It must remain rejected until this phase is complete; Phlex must neither silently reduce a multi-token resource to one token nor emulate multi-token scheduling with a Phlex-specific semaphore.

oneTBB Dependency

The required iterator-range and container constructors for flow::resource_limiter were merged in oneTBB PR #2137. Phlex must wait for an official supported oneTBB release containing those constructors before enabling multi-token resources. Depending on an unreleased upstream branch or commit is out of scope.

Required Work

1. Require an official supported oneTBB release containing the resource_limiter iterator-range or container constructor.
2. Add a CMake feature probe for the required constructor.
3. Materialize tokens() during resource registration and retain token values for the graph lifetime when their type requires it.
4. Construct resource_limiter<T::token_type> from the complete token range or container.
5. Reject an empty token collection at registration time. An empty limiter can leave graph::wait_for_all() blocked.
6. Add tests proving that N tokens allow no more than N concurrent invocations and that multiple token identities are delivered and exercised.

Related Extension: Distinct Single Tokens

A separate extension may support a single token that is not derived from or equal to the registered resource type:

class database {
public:
  using token_type = database_connection*;

  token_type token();
};

The framework would use token() to construct the one-token limiter. This extension does not require runtime-sized oneTBB resource-limiter support and can be implemented independently of multi-token resources.