* fix(chassis): gracefully handle missing NetworkAdapters link (stmcginnis#437)

The Redfish Chassis schema (v1.27.0) defines the NetworkAdapters property
as an optional link to a collection of network adapters physically located
in the chassis. It is perfectly valid for this property to be omitted when
no adapters are present.

However, the current implementation of (*Chassis).NetworkAdapters() calls
ListReferencedNetworkAdapter even when the link is empty, potentially
resulting in 404 errors or misleading failures on otherwise valid systems.

This patch introduces an early return when the networkAdapters link is
not defined, aligning the behavior with other optional properties such as
Power(), Thermal(), and PCIeSlots().

This is particularly relevant for minimal chassis configurations like
backplanes or storage components (e.g., ChassisType: "Component" or "Module"),
which often appear in the Chassis collection but do not expose network
hardware.

Reference: https://redfish.dmtf.org/schemas/v1/Chassis.v1_27_0.json

* Initialize TLSClientConfig when using default transport (stmcginnis#438)

To allow insecure (unverified TLS) connections, we need to set a flag
allowing it. When using the default http client, or if a supplied client
has not been fully configured, the struct containing this flag is not
initialized and set to nil in the http client transport.

This makes sure the config property is initialized if we are configured
to set the flag to avoid nil pointer errors.

Signed-off-by: Sean McGinnis <[email protected]>

* handle etag properly on CreateAccount call (stmcginnis#439)

* Complete support for Thermal.v1_7_1 schema (stmcginnis#440)

Improved handling of fan and temperature sensors data
OEM extensions support for vendor-specific implementations

Co-authored-by: Maxim Kovshov <[email protected]>

* Allow non standard transports to reuse connections. (stmcginnis#441)

* Allow non standard transports to reuse connections.

* Fix whitespace.

* Handle empty arrays for PCIeInterfaces (stmcginnis#443)

* adaptation to the specification https://redfish.dmtf.org/schemas/v1/Power.v1_7_3.json (stmcginnis#445)

added pointers for thresholds as this solves the problem of false triggering of thresholds

Co-authored-by: Maxim Kovshov <[email protected]>

* Refactor Update method to reduce boilerplate (stmcginnis#446)

There is a lot of duplication in the Update calls that doesn't need to
be there. This refactors the update logic to centralize more of the code
and reduce the code duplication.

Signed-off-by: Sean McGinnis <[email protected]>

* adaptation to the specification  https://redfish.dmtf.org/schemas/v1/Sensor.v1_11_0.json (stmcginnis#447)

added pointers for thresholds as this solves the problem of false triggering of thresholds
added tests that show the problem

Co-authored-by: Maxim Kovshov <[email protected]>

* Update PartLocation to handle string int (stmcginnis#448)

Gb200's have LocationOrdinalValue as "3"
This creates a custom unmarshal to handle it.

Co-authored-by: Jimmy Spagnola <[email protected]>

* UP to VirtualMedia.v1_6_5.json (stmcginnis#449)

https://redfish.dmtf.org/schemas/v1/VirtualMedia.v1_6_5.json
once again I solved the problem with the fact that there are so-called imaginary data "" which in fact are absent

Co-authored-by: Maxim Kovshov <[email protected]>

* add WithContext function to client (stmcginnis#452)

* Bump actions/checkout from 4 to 5 (stmcginnis#454)

Bumps [actions/checkout](https://github.com/actions/checkout) from 4 to 5.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](actions/checkout@v4...v5)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-version: '5'
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Add support for ActionInfo to common and redfish.VirtualMedia (stmcginnis#423)

Embed Entity and ODataEtag like other types, same for the Oem field set
it as json.RawMessage instead of string. Looks like this type isn't used
by any other type so this should be ok.

* feat(powersupplyunit): add a RawData() function

* temp fix: use 'any' type for RackOffset field

* add raw data to power distribution metrics (#12)

---------

Signed-off-by: Sean McGinnis <[email protected]>
Signed-off-by: dependabot[bot] <[email protected]>
Co-authored-by: Thomas Jegou <[email protected]>
Co-authored-by: Sean McGinnis <[email protected]>
Co-authored-by: Kovshov Maxim <[email protected]>
Co-authored-by: Maxim Kovshov <[email protected]>
Co-authored-by: spags <[email protected]>
Co-authored-by: Jimmy Spagnola <[email protected]>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: Manuel Mendez <[email protected]>
Co-authored-by: bbrown <[email protected]>