Allow coding shred index to be different than data shred index

[pgarg66]

Problem

First coding shred index must be same as the first data shred index in an FEC block. This prevents an FEC block to have more number of coding shreds than number of data shreds (as that'll cause the last coding shred index to overlap with the data shred index of the next FEC block).

Summary of Changes

Decoupled the coding shred index from data shred index. Also, updated erasure recovery code to use the new indexing scheme.

Reference #7428

[codecov]

Codecov Report

Merging #7438 into master will decrease coverage by 11%.
The diff coverage is 52.7%.

@@            Coverage Diff            @@
##           master   #7438      +/-   ##
=========================================
- Coverage    75.9%   64.9%   -11.1%     
=========================================
  Files         244     245       +1     
  Lines       51675   60518    +8843     
=========================================
+ Hits        39243   39291      +48     
- Misses      12432   21227    +8795

[mvines]

lgtm but @carllin's review would be good too

[pgarg66]

Added fec_set_index to shred common header so that data shreds can identify FEC set they belong to. This will be needed for solving #7450