[AutoML] Allow to use serialized IDataView as an input

ML.NET support at least two types of `IDataView` serializations out of the box - text and binary files.

So I can use [one of two](https://github.com/dotnet/machinelearning/issues/3661#issuecomment-489996753) to prepare my data set for AutoML
```csharp
using (var stream = File.Create(textFileName))
mlContext.Data.SaveAsText(data, stream);

using (var stream = File.Create(binFileName))
mlContext.Data.SaveAsBinary(data, stream);
```

But when I try to use serialized file as an input for AutoML (both CLI and GUI version) it unable to parse them.

### Binary format
Using binary format
> mlnet auto-train --task binary-classification --dataset "data-bin.idv" --label-column-name IsCS --cache on --max-exploration-time 60 --verbosity diag

I see following error
```
Inferring Columns ...
An Error occured during inferring columns
Unable to split the file provided into multiple, consistent columns.
Microsoft.ML.AutoML.InferenceException: Unable to split the file provided into multiple, consistent columns.
at Microsoft.ML.AutoML.ColumnInferenceApi.InferSplit(MLContext context, TextFileSample sample, Nullable`1 separatorChar, Nullable`1 allowQuotedStrings, Nullable`1 supportSparse)
at Microsoft.ML.AutoML.ColumnInferenceApi.InferColumns(MLContext context, String path, ColumnInformation columnInfo, Nullable`1 separatorChar, Nullable`1 allowQuotedStrings, Nullable`1 supportSparse, Boolean trimWhitespace, Boolean groupColumns)
at Microsoft.ML.CLI.CodeGenerator.AutoMLEngine.InferColumns(MLContext context, ColumnInformation columnInformation)
at Microsoft.ML.CLI.CodeGenerator.CodeGenerationHelper.GenerateCode()
at Microsoft.ML.CLI.Program.<>c__DisplayClass1_0.b__0(NewCommandSettings options)
Please see the log file for more info.
Exiting ...
```

### Text format

With `--verbosity diag` it stuck on the line
```
Inferring Columns ...
Creating Data loader ...
Loading data ...
Exploring multiple ML algorithms and settings to find you the best model for ML task: binary-classification
For further learning check: https://aka.ms/mlnet-cli
| Trainer Accuracy AUC AUPRC F1-score Duration #Iteration |
[Source=AutoML, Kind=Trace] Channel started
```

with default verbosity
> mlnet auto-train --task binary-classification --dataset "data-txt.tsv" --label-column-name IsCS --cache on --max-exploration-time 60

it return an error of type mismatch
```
xploring multiple ML algorithms and settings to find you the best model for ML task: binary-classification
For further learning check: https://aka.ms/mlnet-cli
──────────────────────────
Waiting for the first iteration to complete ... 00:00:00
Exception occured while exploring pipelines:
Provided label column 'IsCS' was of type Single, but only type Boolean is allowed.
Please see the log file for more info.
```

but data file looks correct (it serialized by ML.NET).
This is the header and first lines of dataset
```
#@ TextLoader{
#@ header+
#@ sep=tab
#@ col=IsCS:BL:0
#@ col=Features:R4:1-19
#@ }
IsCS 19 0:""
0 2 0.259255171 0 0 0 1.41421354 0 1.41421354 0 1.41421354 0 1.41421354 0 3 6 0 0 1 1192
0 6 0.259255171 0 0 0 1.41421354 0 1.41421354 0 1.41421354 0 1.41421354 0 3 6 0 0 1 1192
```