Large integers should be read as strings, not as numbers

**What's wrong?**
I have a list of user_IDs (often 17-digit numbers, but also shorter ones). After data import through the Import Document widget, some of these "longer" user IDs get changed with regard to the original IDs that are listed in the imported file.

Rendition in Orange :

Orange -----> TSV data imported -----> Timestamp ----->Note
10213407285517228----->10213407285517229----->2017-06-25T17:37:17+0000 -----> in Orange: DIFFERS from the imported data
10155206846088648----->10155206846088649----->2017-06-25T17:45:31+0000 -----> in Orange: DIFFERS from the imported data
2049903568368818----->2049903568368818----->2014-01-08T09:03:21+0000 -----> in Orange: IDENTICAL to the imported data
10211717072377002----->10211717072377002----->2014-01-07T16:56:31+0000 -----> in Orange: IDENTICAL to the imported data

**What's your environment?**

- Operating system: Windows
- Orange version: 3.33.0

Start by reproducing the mismatch with the TSV examples in the Import Document widget. Check how the widget interprets the listed user IDs during import; done means the imported values remain identical to the source, including the 17-digit IDs.

Written by the indexing model from the issue text.