Decode error correction code with dictionary

[tim-tx]

In the note at the top of a PDF output indicating what qr-backup command you used to generate the document, the reported error correction code is wrong. For example, use the default M and the report will say you used --error-correction L. The reported value gets decoded by indexing the string LMQH but the qrcode module actually assigns

ERROR_CORRECT_L = 1
ERROR_CORRECT_M = 0
ERROR_CORRECT_Q = 3
ERROR_CORRECT_H = 2

see https://github.com/lincolnloop/python-qrcode/blob/main/qrcode/constants.py

The PR just decodes from a dictionary (i.e. by name) in case those values were to change.

[tim-tx]

Tests are failing because they compare hashes of PDF outputs to pre-calculated values. The PDF hashes could either be updated in the tests or the tests could be re-written to just compare restored data with original data.