Optimize data section parsing and storage

[Robbepop]

Benchmarks revealed that parsing and handling of data sections were a big contributor to the overall spent time for some benchmarks, such as ffmpeg.wasm. This is probably due to heap memory allocator pressure since all of its many data segments each allocated a bit of heap memory for their bytes.
This PR bundles bytes of all active data segments into a single buffer reducing the overall memory consumption and pressure on the heap memory allocator.
Benchmarks show roughly 15% improvement for compiling ffmpeg.wasm in lazy compilation mode.

[codecov]

Codecov Report

Attention: Patch coverage is 75.26882% with 23 lines in your changes are missing coverage. Please review.

Project coverage is 80.77%. Comparing base (30ca773) to head (957cc9f).

Files	Patch %	Lines
crates/wasmi/src/module/data.rs	80.76%	10 Missing ⚠️
crates/wasmi/src/memory/data.rs	53.33%	7 Missing ⚠️
crates/wasmi/src/module/instantiate/mod.rs	66.66%	5 Missing ⚠️
crates/wasmi/src/module/builder.rs	87.50%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1024      +/-   ##
==========================================
- Coverage   80.80%   80.77%   -0.03%     
==========================================
  Files         269      269              
  Lines       24736    24786      +50     
==========================================
+ Hits        19989    20022      +33     
- Misses       4747     4764      +17     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.