[Bug] Not preserving zero padding

**Describe the bug**
Type analysis in kepler.gl doesn't preserve zero padded fields, which are sometimes important, such as zip codes. Also, the type analyzer sometimes detects things as time, perhaps based on the magnitude of the number, but this may or may not be right depending on specifics of the data.

**To Reproduce**
Use the following CSV file, containing an integer, a census block group ID, and a zero padded integer:
```
a,b,d
36,360610199004,06
```

I get the fields analyzed as `int`, `time`, and `int`.

**Expected behavior**

The second field might be a timestamp, unclear from this example.
The last field is analyzed as a string.

Start with kepler.gl's type analysis using the CSV example in this issue and compare how the fields `36`, `360610199004`, and `06` are classified. Done means the zero-padded `06` is analyzed as a string; the expected classification of the census block group ID remains unresolved in the issue.

Written by the indexing model from the issue text.