Handle `allOf` types more gracefully

[karencfv]

Currently as the OpenAPI spec we use treats allOf types as enums, so they are being handled that way. Ideally it'd be best to treat them as allOf types.

Update: After #78 these values are treated as interfaces. This way you can pass whichever
of the struct types you need like this:

ipRange := oxide.Ipv4Range{
	First: "172.20.15.240",
	Last:  "172.20.15.250",
}

body := oxide.IpRange(ipRange)
resp, _ := client.IpPoolRangeAdd("mypool", &body)

Probably not the best approach, but will leave them this way until I come up with
a more idiomatic solution. Keep an eye out on this one to refine.

[karencfv]

Looking into this it looks like it may not be possible to use generic types for this.

I was looking into implementing in this manner:

type IpRange interface{
	Ipv4Range | Ipv6Range
}

type Ipv4Range struct {
	First string `json:"first,omitempty" yaml:"first,omitempty"`
	Last  string `json:"last,omitempty" yaml:"last,omitempty"`
}

type Ipv6Range struct {
	First string `json:"first,omitempty" yaml:"first,omitempty"`
	Last  string `json:"last,omitempty" yaml:"last,omitempty"`
}

type IpPoolRange[T IpRange] struct {
	Id          string     `json:"id,omitempty" yaml:"id,omitempty"`
	Range       T    `json:"range,omitempty" yaml:"range,omitempty"`
	TimeCreated *time.Time `json:"time_created,omitempty" yaml:"time_created,omitempty"`
}

type IpPoolRangeResultsPage[T IpRange] struct {
	Items []IpPoolRange[T] `json:"items,omitempty" yaml:"items,omitempty"`
	NextPage string `json:"next_page,omitempty" yaml:"next_page,omitempty"`
}

func (c *Client) IpPoolServiceRangeAdd[T IpRange](rackId string, j *T) (*IpPoolRange[T], error) {
  ...
}

Unfortunately I hit a roadblock where apparently you can't use use generics with methods (only functions). There is some discussion on the matter.

This is really unfortunate. I may have to resort to type checking which is a real shame 😞

[sudomateo]

We're discussing using a similar interface approach for oneOf enums in #342 and on RFD 0614. Going to close this issue for now since we're okay with the current state of allOf types.