Feature/streaming encryption

[jbygdell]

Related issue(s) and PR(s)
This PR closes #473 .

Description
This PR creates a streaming encryption function that is used when uploading files with the -encrypt-with-key flag set. The files are encrypted in memory during the upload process using a 1MB blocking ring buffer.
The ring buffer will block reading of data if it is empty and will at most contain 1MB of data during the upload process in order not to consume all system memory when uploading large files. Reads and writes are matched to the buffer once it is full i.e. for each chunk that is read one can be written to the buffer.

The PR also contains a bunch of linter fixes that needed to be fixed.

How to test

[pahatz]

The encryption itself and the upload seem to work great!
I do have some concerns though.

1. The checksums that we produce will be difficult to use in the current naming format.
   For example, I ran a recursive upload of a folder with two files and this produces the following files in the directory of the executable:

-rw-------. 1 pan  pan      1111 Jul 23 12:55 checksum_unencrypted.sha256
-rw-------. 1 pan  pan       663 Jul 23 12:55 checksum_unencrypted.md5
-rw-------. 1 pan  pan      1221 Jul 23 12:55 checksum_encrypted.sha256
-rw-------. 1 pan  pan       773 Jul 23 12:55 checksum_encrypted.md5

There should be some naming convention to be able to link them to the original and maybe place them in the same directory.

2. The progress bar seems to be broken with this implementation.

[jbygdell]

1. The checksums that we produce will be difficult to use in the current naming format.
   For example, I ran a recursive upload of a folder with some files and this produces the following files in the directory of the executable:

Nothing has changed there with regards to how it currently is done.
The files contain the relevant information i.e. hash and file name.

2. The progress bar seems to be broken with this implementation.
   In what sense is it broken?

I get the same visual appearance when uploading both for encrypted and unencrypted:

File README.md: uploading 0.0 b / 20.2 KiB [------]

[jbygdell]

@pahatz I think I managed to fix the visuals now.

[pahatz]

Fair enough, I hadn't checked before how our checksums look.
The progress bar looks good now, great work!

[nanjiangshu]

Looks good in general. But I have two comments, the first one must be fixed and the second one is better to fixed.

1. The uploaded c4gh files have only 124 Bytes, that is, only with the header, when the files to be uploaded with -encrypt-with-key is less than 1Mb.

2. When there are encrypted version of files in the folder, e.g. in the following example, I have a folder called test_dir2 and it contains both file_100.txt and file_100.txt.c4gh. The command exit with the error Error: aborting, file is already encrypted.

It is better to show also the name of the file that is already encrypted is file_100.txt.c4gh. Otherwise, it looks like the process is aborted because of file_100.txt.c4gh` but it's not.

Previously files with .c4gh extension will be ignored, but now any files in the folder will be processed, even those with .c4gh extension. It is better to point out the exact file that caused the abortion.

$ ../sda-cli -config s3cmd-bp-staging.conf upload -encrypt-with-key mykey.pub.pem  -r test_dir2  -targetDir test_dir2_try2
Remote server (host_base): https://staging-inbox.bp.nbis.se
File file_100.txt: uploading 124.0 b / 292.0 b [========================>-----------------------------------]  42 %
Error: aborting, file is already encrypted

[jbygdell]

2. When there are encrypted version of files in the folder, e.g. in the following example, I have a folder called test_dir2 and it contains both file_100.txt and file_100.txt.c4gh. The command exit with the error Error: aborting, file is already encrypted.

While not part of the change originally it is easy to fix.

[jbygdell]

1. The uploaded c4gh files have only 124 Bytes, that is, only with the header, when the files to be uploaded with -encrypt-with-key is less than 1Mb.

Related to an unclosed reader/Writer. Extra tests has been added

[nanjiangshu]

Looks good!