fix(net): skip v1 physical entries with no name

[shAsh-cy]

• I have signed the CLA: https://ubuntu.com/legal/contributors
• I have included a comprehensive commit message using the guide below
• I have added unit tests to cover the new behavior under tests/unittests/
• I have kept the change small, avoiding unnecessary whitespace or non-functional changes.
• I have added a reference to issues that this PR relates to in the PR message
• I have updated the documentation with the changed behavior. Not applicable, no user-facing change.

Proposed Commit Message

fix(net): skip v1 physical entries with no name

_version_1 skipped entries without a mac_address but not entries
without a name, so a None name reached device_driver() and raised
TypeError in sys_dev_path(). This crashed wait_for_physdevs while
applying network config. Skip nameless entries, matching the
existing behaviour of _version_2.

This uses the same `if not name` condition as _version_2, so an
entry with an empty name is skipped too. That case does not crash
today, but an empty name is equally unusable for renaming.

Fixes GH-6982

Additional Context

The crash. A v1 entry of type physical with a mac_address but no usable name reaches device_driver(name) with name=None, which raises TypeError in sys_dev_path(). That propagates out of wait_for_physdevs() in Init.apply_network_config(), which is not wrapped, so network config fails. DataSourceHetzner emits such an entry when a private NIC has not hotplugged yet. Note that device_devid(name) on the following line has the same exposure, so a guard around only device_driver would have moved the crash rather than fixed it.

Why skip rather than guard the lookups. I first considered guarding only the device_driver/device_devid calls and letting the row through with name=None, to keep the MAC in wait_for_physdevs's expected_macs. Tracing the other consumer showed that is worse. extract_physdevs also feeds _rename_interfaces() via apply_network_config_names(). With driver and device_id both None, entry_match() falls through to matching on MAC alone, so a nameless row matches a present device and reaches ("rename", mac, None, (cur_name, None)). If that device is up and downable it is taken down first, the rename then fails, and the bring-up fails too, leaving the interface down. _apply_netcfg_names() wraps all of this in except Exception and logs a warning, so it would fail silently. Today that path is unreachable because the TypeError fires first. Skipping in _version_1 keeps the row out of both consumers.

Why this cannot regress. Every v1 config reaching this path with a nameless physical entry crashes today, so there is no working behaviour for the skip to change.

Trade-off. The MAC is no longer waited for in wait_for_physdevs. If nameless-but-MAC'd v1 entries are considered legitimate input worth waiting for, the alternative is to keep the row and reject it in _rename_interfaces instead. Happy to do that as a follow-up if preferred.

Test Steps

Before this change:

>>> from cloudinit.net import extract_physdevs
>>> extract_physdevs({"version": 1, "config": [{"type": "physical",
...   "mac_address": "86:00:00:3c:5c:43", "name": None,
...   "subnets": [{"type": "dhcp", "ipv4": True}]}]})
TypeError: can only concatenate str (not "NoneType") to str

After this change the same call returns []. A well-formed v1 entry is unaffected and still resolves to [['86:00:00:3c:5c:43', 'eth0', 'virtio', '0x1000']].

New unit test: TestExtractPhysdevs::test_get_v1_type_physical_skips_if_no_name, placed alongside the existing v2 equivalent. It asserts the entry is skipped and that device_driver/device_devid are never called. The latter matters because the class's autouse fixture mocks both, so the return-value assertion alone would not pin the actual defect.

Rebased onto bb81cd0. Full suite: 5744 passed, 5 skipped, 13 xfailed (tox -e py3). tox -e check_format passes.

Merge type

• Squash merge using "Proposed Commit Message"
• Rebase and merge unique commits. Requires commit messages per-commit each referencing the pull request number (#<PR_NUM>)