Generated two large test files and removed

[mattdowle]

Placed R code from issue #2275 into tests.Rraw to auto generate DT1 and DT2 rather than fread them in. Removed the need for stringi. The random data isn't exactly the same since stringi uses a different RNG I assume, but I don't think the particular seed matters in this case. Please check though.
To pass CRAN check note :

* checking installed package size ... NOTE
  installed size is  9.3Mb
  sub-directories of 1Mb or more:
    libs    1.1Mb
    tests   6.6Mb

due to these files from PR #2461 :

$ ls -lh inst/tests/issue_2275_dt*.csv
-rw-r--r-- 1 mdowle mdowle 1.7M Nov  8 12:31 inst/tests/issue_2275_dt1.csv
-rw-r--r-- 1 mdowle mdowle 2.7M Nov  8 12:31 inst/tests/issue_2275_dt2.csv

[codecov-io]

Codecov Report

Merging #2471 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #2471   +/-   ##
=======================================
  Coverage   91.59%   91.59%           
=======================================
  Files          63       63           
  Lines       12031    12031           
=======================================
  Hits        11020    11020           
  Misses       1011     1011

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1b8e886...177e5ad. Read the comment docs.

[arunsrinivasan]

Thanks, Matt. I tried to reduce the size, but couldn't replicate the issue. I didn't realise the file was too big. I tested this data with the commit prior to the commit that fixed it and am able to reproduce the issue. So all good.