[crashtracker] Gate compilation using cargo features

[danielsn]

What does this PR do?

Adds two new cargo features: collector and receiver. Refactors the code to make it easy to use those cfg to define what gets built at compile time.

Motivation

The current implementation of the library always builds in all code. This is not great, because it bloats binary size etc. We have users who

1. (.Net) independently collect core-dumps, and then format them into CrashInfo objects, which they upload. They have no need or use for the receiver or collector portions of this code
2. (Ruby, Python) Use the collector to gather info on crashes, and the receiver to format a CrashInfo object and send it on. They need all parts of the code.
3. (PHP). Propose to collect the info on crashes in the main binary, then use a sidecar to format a CrashInfo object and send it on. The first binary needs the collector but not the receiver; the sidecar needs the opposite.

Additional Notes

• Currently, we don't have support when building the ffi library for feature selection. This is a first step towards making that worthwhile.

• I did some refactoring as I moved code around, but there is still more refactoring that could be done.

How to test the change?

This is a refactor, existing tests cover the functionality

PROF-10245

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 4.59184% with 187 lines in your changes missing coverage. Please review.

Project coverage is 70.45%. Comparing base (89d4831) to head (6da7160).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #541      +/-   ##
==========================================
- Coverage   70.45%   70.45%   -0.01%     
==========================================
  Files         208      213       +5     
  Lines       28414    28408       -6     
==========================================
- Hits        20020    20014       -6     
  Misses       8394     8394              

Components	Coverage Δ
crashtracker	25.21% <5.68%> (-0.33%)	⬇️
datadog-alloc	98.73% <ø> (ø)
data-pipeline	50.00% <ø> (ø)
data-pipeline-ffi	0.00% <ø> (ø)
ddcommon	86.43% <ø> (ø)
ddcommon-ffi	75.29% <ø> (ø)
ddtelemetry	58.95% <ø> (ø)
ipc	84.13% <ø> (ø)
profiling	78.09% <3.70%> (ø)
profiling-ffi	56.76% <3.70%> (ø)
serverless	0.00% <ø> (ø)
sidecar	35.42% <ø> (ø)
sidecar-ffi	0.00% <ø> (ø)
spawn-worker	54.98% <ø> (ø)
trace-mini-agent	71.14% <ø> (ø)
trace-normalization	98.24% <ø> (ø)
trace-obfuscation	95.73% <ø> (ø)
trace-protobuf	77.16% <ø> (ø)
trace-utils	91.03% <ø> (ø)

[pr-commenter]

Benchmarks

This comment was omitted because it was over 65536 characters.Please check the Gitlab Job logs to see its output.

Looks like this also needs changes to ffi

[sanchda]

LGTM--pretty reasonable-looking refactor with minimal changes to how the code operates.

[sanchda]

Silly question: do these copyright notices indicate the year of first existence for the noted file, or the last update?

[danielsn]

I assume first year of existence. At Amazon we stopped using years, and went to a SPDX header which was much cleaner