Better explain identifiers

Ok, we have some schema:

```
cwlVersion: v1.0
class: CommandLineTool
baseCommand: [tar, xf]
inputs:
tarfile:
type: File
inputBinding:
position: 1
tarfile.path:
type: string
inputBinding:
position: 2
outputs:
example_out:
type: File
outputBinding:
glob: $(inputs.tarfile.path)
```

and values file:

```
tarfile:
class: File
path: hello.tar
tarfile.path: goodbye.txt
```

What value will be in glob field? tarfile.path as input or path field in tarfile?

So, I have two questions:

1. What standard says about this situation? I've not found answer in documentation;
2. Is it correct situation? Because, we don't have access to the second input at all or field in first input;

Thanks!

Use the supplied CWL v1.0 schema and values file as the reproduction case. Start by checking the CWL v1.0 specification and existing documentation for identifier resolution and field access. Done means documenting which tarfile.path interpretation applies, whether the example is valid, and how the two questions should be answered.

Written by the indexing model from the issue text.