CWLtool issue with passing a list of scatter/gathered outputs into a program that accepts a list

This works:

```
#!/usr/bin/env cwl-runner
class: Workflow
requirements:
- class: ScatterFeatureRequirement
inputs:
- id: "#mutation"
type: string
- id: "#normalin"
type: File
- id: "#tumorin"
type: File
outputs:
- id: "#outfile"
type: File
source: "#wc.outfile"
steps:
- id: "#greptumor"
run: {import: grep.cwl.yaml}
#scatter: "#grep.infile"
inputs:
- id: "#grep.pattern"
source: "#mutation"
- id: "#grep.infile"
source: "#tumorin"
outputs:
- id: "#greptumor.outfile"
- id: "#grepnormal"
run: {import: grep.cwl.yaml}
#scatter: "#grep.infile"
inputs:
- id: "#grep.pattern"
source: "#mutation"
- id: "#grep.infile"
source: "#normalin"
outputs:
- id: "#grepnormal.outfile"
- id: "#wc"
run: {import: wc.cwl.yaml}
inputs:
- id: "#wc.infile"
source: ["#grepnormal.outfile", "#greptumor.outfile"]
outputs:
- id: "#wc.outfile"
```

This does not work, when I try to have it work on multiple normal/tumor input files into grep scattered:

```
#!/usr/bin/env cwl-runner
class: Workflow
requirements:
- class: ScatterFeatureRequirement
inputs:
- id: "#mutation"
type: string
- id: "#normalin"
type: {type: array, items: File}
- id: "#tumorin"
type: {type: array, items: File}
outputs:
- id: "#outfile"
type: File
source: "#wc.outfile"
steps:
- id: "#greptumor"
run: {import: grep.cwl.yaml}
scatter: "#grep.infile"
inputs:
- id: "#grep.pattern"
source: "#mutation"
- id: "#grep.infile"
source: "#tumorin"
outputs:
- id: "#greptumor.outfile"
- id: "#grepnormal"
run: {import: grep.cwl.yaml}
scatter: "#grep.infile"
inputs:
- id: "#grep.pattern"
source: "#mutation"
- id: "#grep.infile"
source: "#normalin"
outputs:
- id: "#grepnormal.outfile"
- id: "#wc"
run: {import: wc.cwl.yaml}
inputs:
- id: "#wc.infile"
source: ["#grepnormal.outfile", "#greptumor.outfile"]
outputs:
- id: "#wc.outfile"
```

Interestingly though, this small change does work, when you only pass one of the scatter/gathered inputs into the final wc.

```
#!/usr/bin/env cwl-runner
class: Workflow
requirements:
- class: ScatterFeatureRequirement
inputs:
- id: "#mutation"
type: string
- id: "#normalin"
type: {type: array, items: File}
- id: "#tumorin"
type: {type: array, items: File}
outputs:
- id: "#outfile"
type: File
source: "#wc.outfile"
steps:
- id: "#greptumor"
run: {import: grep.cwl.yaml}
scatter: "#grep.infile"
inputs:
- id: "#grep.pattern"
source: "#mutation"
- id: "#grep.infile"
source: "#tumorin"
outputs:
- id: "#greptumor.outfile"
- id: "#grepnormal"
run: {import: grep.cwl.yaml}
scatter: "#grep.infile"
inputs:
- id: "#grep.pattern"
source: "#mutation"
- id: "#grep.infile"
source: "#normalin"
outputs:
- id: "#grepnormal.outfile"
- id: "#wc"
run: {import: wc.cwl.yaml}
inputs:
- id: "#wc.infile"
source: ["#grepnormal.outfile"]
outputs:
- id: "#wc.outfile"
```

So it seems to me that this is an issue only when you use the program accepting an array of inputs and you want to pass two of those scatter/gathered things into a final output. Maybe the data structure being output in the scatter/gather is weird when you put it into a list? The error is this about a dict being expected:

```
> cwltool mutationfinder3.cwl.yaml --mutation GCATCCA --normalin normal.fastq --tumorin tumor.fastq
/anaconda/bin/cwltool 1.0.20151026181844
[job 4386765520] /var/folders/8r/g00jq11j2yb586cz04tfpydc0000gn/T/tmpm8fJ5E$ grep GCATCCA /Users/john/workspaces/commonwl_examples/t790m_detector/normal.fastq > /var/folders/8r/g00jq11j2yb586cz04tfpydc0000gn/T/tmpm8fJ5E/out.txt
[job 4386765840] /var/folders/8r/g00jq11j2yb586cz04tfpydc0000gn/T/tmpUclOsb$ grep GCATCCA /Users/john/workspaces/commonwl_examples/t790m_detector/tumor.fastq > /var/folders/8r/g00jq11j2yb586cz04tfpydc0000gn/T/tmpUclOsb/out.txt
Workflow error:
Error validating input record, could not validate field `infile` because
At position 0
`[{'checksum': 'sha1$e62341a47da9cb155ec0f531b525a97d2bd86455',
'class': 'File',
'path': '/var/folders/8r/g00jq11j2yb586cz04tfpydc0000gn/T/tmpm8fJ5E/out.txt'[...]`
is not a dict
```