Refactor: Extract shared key cache into separate file

[sudden6]

This removes the old and badly documented/designed shared key cache from DHT.c and moves the implemenation to a separate file. This makes it more testable and allows a better overview of how many caches exist (currently 7, total ~0,5MB). Additionally I think this fixes several missed crypto_memzero calls.

---

This change is 

[codecov]

Codecov Report

Merging #2317 (de97532) into master (e0b00d3) will increase coverage by 0.09%.
The diff coverage is 91.73%.

@@            Coverage Diff             @@
##           master    #2317      +/-   ##
==========================================
+ Coverage   78.08%   78.17%   +0.09%     
==========================================
  Files         140      141       +1     
  Lines       31053    31097      +44     
==========================================
+ Hits        24247    24311      +64     
+ Misses       6806     6786      -20     

Impacted Files	Coverage Δ
toxcore/announce.c	58.80% <66.66%> (+0.41%)	⬆️
toxcore/onion.c	85.71% <87.50%> (+0.60%)	⬆️
toxcore/onion_announce.c	85.76% <87.50%> (+0.31%)	⬆️
toxcore/shared_key_cache.c	94.64% <94.64%> (ø)
toxcore/DHT.c	84.74% <100.00%> (-0.34%)	⬇️
toxcore/net_crypto.c	87.37% <100.00%> (+0.01%)	⬆️
toxcore/ping.c	86.92% <100.00%> (+0.50%)	⬆️
auto_tests/group_invite_test.c	98.44% <0.00%> (-0.78%)	⬇️
toxav/toxav.c	69.08% <0.00%> (-0.15%)	⬇️
... and 11 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[JFreegman]

Reviewed 13 of 14 files at r1, all commit messages.
Reviewable status: 0 of 1 approvals obtained (waiting on @sudden6)

---

toxcore/announce.c line 438 at r1 (raw file):

    VLA(uint8_t, plain, plain_len);

    const uint8_t* shared_key = shared_key_cache_lookup(announce->shared_keys, data_public_key);

shared_key_cache_lookup() can return null. Wouldn't it be better to do a null check on the return value immediately and log any errors, instead of propagating the null pointer to decrypt_data_symmetric() and obfuscating the error?

---

toxcore/shared_key_cache.h line 20 at r1 (raw file):

/**
 * @brief Initializes a new shared key cache

Missing period (a couple more in this file)

[sudden6]

Reviewable status: 0 of 1 approvals obtained (waiting on @JFreegman)

---

toxcore/announce.c line 438 at r1 (raw file):

Previously, JFreegman wrote…

shared_key_cache_lookup() can return null. Wouldn't it be better to do a null check on the return value immediately and log any errors, instead of propagating the null pointer to decrypt_data_symmetric() and obfuscating the error?

Done.

---

toxcore/shared_key_cache.h line 20 at r1 (raw file):

Previously, JFreegman wrote…

Missing period (a couple more in this file)

Done.

[JFreegman]

Reviewed 5 of 5 files at r2, all commit messages.
Reviewable status: complete! 1 of 1 approvals obtained