Schema conversion conflates "optional field" with "allowing null values"

Similar issue was reported [here](https://github.com/instrumenta/kubernetes-json-schema/issues/25) but this might be a better home for it, not sure.

In [this code](https://github.com/instrumenta/openapi2jsonschema/blob/master/openapi2jsonschema/util.py#L69) the tool conflates an optional field (i.e. a field that does not have to be specified) with a field that allows ``null`` values. But this is not correct and these are distinct things in some cases (e.g. Kubernetes schemas)

As a real world example, if you have a Kubernetes ``Deployment`` and set a secret volume like this (with ``secretName`` having the ``null`` value)...

```yaml
...
spec:
volumes:
- name: somevol
secret:
secretName:
...
```

The schema produced from ``openapi2jsonschema`` using ``kubeval`` will happily consider this config valid. But if you try to apply the config using ``kubectl`` or ``kustomize``, it complains that ``secretName`` cannot take a null value. And indeed if you look at the upstream schema, nowhere does it specify that the field is nullable.