stop Gson exposure by making MastodonClient.getSerializer internal

[bocops]

This closes #51.

There are some weird things going on that I want to mention here:

• mocking an otherwise inaccessible function via dynamic call in the MockClient for bigbone-rx. This might be a sign of either a badly designed test, or other structural issues.
• creating separate Gson() instances directly in our samples. This should typically not be done, but those samples use undocumented functionality and need to be rewritten anyway.

Neither is really great, but we shouldn't keep MastodonClient.getSerializer() public only to keep tests and outdated samples working.

I also had to use @PublishedApi on the now internal function. That means that the function is "technically exposed in bytecode" but not able to be called directly from outside the module. I assume this is OK.

[codecov]

Codecov Report

Merging #86 (c9e77c6) into master (ae9691f) will not change coverage.
The diff coverage is 0.00%.

Additional details and impacted files

@@            Coverage Diff            @@
##             master      #86   +/-   ##
=========================================
  Coverage     41.72%   41.72%           
  Complexity      167      167           
=========================================
  Files            62       62           
  Lines          1383     1383           
  Branches         82       82           
=========================================
  Hits            577      577           
  Misses          785      785           
  Partials         21       21           

Impacted Files	Coverage Δ
...e/src/main/kotlin/social/bigbone/MastodonClient.kt	10.13% <0.00%> (ø)

[andregasser]

All good for me. Good to have getSerializer() no longer exposed to library users.