refactor TextToPDF.call method

[valerybokov]

Current algorithm is:
1 if charset is UTF8 then read 3 bytes.
2 if these 3 bytes have expected values then mark a hasUtf8BOM variable as true
3 close stream
4 open new stream of the same file
5 if the variable hasUtf8BOM is true then skip 3 bytes.
6 if couldn't skip 3 bytes then throw an exception
7 If bytes were skipped or there is no need to skip them, the rest of the file should be read.

These are questions not for you, but for the algorithm:
1 why we need read the file twice when it increases the likelihood that we won't succeed the second time?
2 Opening a stream twice is slower than opening it once.
3 According to the code, there's a possibility that we couldn't read the file a second time. Then why isn't there a check to see if the file is corrupted? That is, it's UTF-8 encoding, but what if one or two of these three bytes are different. Perhaps the format has such combinations that this can't be verified.

At this point, I propose making one stream instead of two. There are no other changes.