b14f2c7 tests: applied PYTHON_GIL to the env for every test (kevkevinpal)

Pull request description:

  ## Summay
  If a user is running python3.14.0t they would see a warning log that would fail the integration test suite.

  This change adds `PYTHON_GIL=1` to the env when running our functional test suite to ensure that the tests pass for users running python3.14.0t and are not manually setting `PYTHON_GIL=1`.

  This resolves #33582

  ### Tests before and after
  #### Before
  ```
  ./build/test/functional/test_runner.py interface_ipc.py
  Temporary test directory at /tmp/test_runner_₿_🏃_20260319_142327
  Remaining jobs: [interface_ipc.py]
  1/1 - interface_ipc.py failed, Duration: 2 s

  stdout:
  2026-03-19T18:23:27.330123Z TestFramework (INFO): PRNG seed is: 4933091336597497631
  2026-03-19T18:23:27.380917Z TestFramework (INFO): Initializing test directory /tmp/test_runner_₿_🏃_20260319_142327/interface_ipc_0
  2026-03-19T18:23:28.625944Z TestFramework (INFO): Running echo test
  2026-03-19T18:23:28.635856Z TestFramework (INFO): Running mining test
  2026-03-19T18:23:28.648965Z TestFramework (INFO): Running deprecated mining interface test
  2026-03-19T18:23:28.653589Z TestFramework (INFO): Running disconnect during BlockTemplate.waitNext
  2026-03-19T18:23:28.821124Z TestFramework (INFO): Running thread busy test
  2026-03-19T18:23:29.195589Z TestFramework (INFO): Stopping nodes
  2026-03-19T18:23:29.299135Z TestFramework (INFO): Cleaning up /tmp/test_runner_₿_🏃_20260319_142327/interface_ipc_0 on exit
  2026-03-19T18:23:29.299329Z TestFramework (INFO): Tests successful

  stderr:
  <frozen importlib._bootstrap>:491: RuntimeWarning: The global interpreter lock (GIL) has been enabled to load module 'capnp.lib.capnp', which has not declared that it can run safely without the GIL. To override this behavior and keep the GIL disabled (at your own risk), run with PYTHON_GIL=0 or -Xgil=0.

  TEST             | STATUS    | DURATION

  interface_ipc.py | ✖ Failed  | 2 s

  ALL              | ✖ Failed  | 2 s (accumulated)
  Runtime: 2 s

  ```
  #### After
  ```
  ./build/test/functional/test_runner.py interface_ipc.py
  Temporary test directory at /tmp/test_runner_₿_🏃_20260319_142221
  Remaining jobs: [interface_ipc.py]
  1/1 - interface_ipc.py passed, Duration: 2 s

  TEST             | STATUS    | DURATION

  interface_ipc.py | ✓ Passed  | 2 s

  ALL              | ✓ Passed  | 2 s (accumulated)
  Runtime: 2 s
  ```

ACKs for top commit:
  maflcko:
    review ACK b14f2c7
  fanquake:
    ACK b14f2c7

Tree-SHA512: e5862d2e9211154d4834c88864e8c4e35de195986511ba151871d39266d177e0718960b28020e815ef6b353a0d82800b7cb68e9a6dee82fc85f12d8705e787a8

This refactor makes the field a bit more type-safe.

This also better reflects the meaning and use.

…mestamp

faaea78 refactor: Use current_time over redundant call to Now() (MarcoFalke)
3333c50 refactor: Use NodeClock::time_point for m_addr_token_timestamp (MarcoFalke)
fa55723 move-only: Extract ProcessAddrs() helper (MarcoFalke)

Pull request description:

  It is a bit confusing to have some code use the deprecated `GetTime`, which returns a duration and not a time point, and other code to use `NodeClock` time points.

  Fix one place `m_addr_token_timestamp` to use `NodeClock::time_point`.

  Also:

  * Extract a `ProcessAddrs` helper, similar to the other `Process*()` helpers, to cut down the `ProcessMessage` with a massive scope.
  * Rename the confusing `current_a_time` to `now_seconds`. (The `a` in this context refers to the removed "adjusted" time, see commit fadd8b2, which removed adjusted time here)

ACKs for top commit:
  l0rinc:
    ACK faaea78
  ajtowns:
    reACK faaea78
  sedited:
    Re-ACK faaea78

Tree-SHA512: 67ad13e9d7b88e08e3d723e6b7cd598b38df2a004f5c2338b24f2992e25ae9d8fb8e5325c9c94171e551fe86d87e3e3ec1fe6baae64edbf6b5c125f408ee64e4