Fix tests and docs after introducing possibility of `indicator=nothing`

[JanJereczek]

This PR mainly fixes the tests and docs that broke after the introduction of passing indicators = nothing (#65).

Majors:

• When introducing the possibility of passing nothing as an indicator, the need of complementing an if-statement in significant_transitions was overseen. This is now corrected such that the surrogate is correctly passed to the analysis pipeline regardless of the case.
• "nothing indicator" in full_analysis.jl needs @test count(flags) == 2 since both indicators are positive at the transition.
• The previous PR requires the number of change metrics to correspond to the number of indicators. This was corrected across all docs files.
• Modification suggestion: let us require the tail to be of the same size as indicators and change_metrics too. This would be more consistent and easier to maintain. So far, I applied this to the surrogate analysis. If you agree, we can make the same change across all types of significance analysis.

Minors:

• Updated compat for Documenter.jl s.t. >= 1. Note that with this version, I had to use warnonly = true since some referencing were not found. I tried to solve this, which only partially succeeded (I don't understand why the ref to e.g. PrecomputedLowfreqPowerSpectrum is not found).
• The docs of Statistics.jl has many refs that were not found. To prevent this, I rather used the docs of StatsBase.jl, which are equally valid for that matter. I am sure there is a more elegant solution to that though. However, this makes me ask: we are we not simply using everything from StatsBase.jl? For instance also mean and var?
• Updated compat for TimeseriesSurrogates.jl s.t. version with fixed memory allocation for Fourier surrogates is used.
• In docs/make.jl, I simplified loading of example files. Before that we were loading by reading the directory, looping over it and then permute the vector to have the desired order. This is actually simpler to just write manually.
• Shifted difference_of_maxes to src/ for potential future use.
• Small corrections across documentation...

[codecov-commenter]

Codecov Report

Attention: 69 lines in your changes are missing coverage. Please review.

Comparison is base (971564b) 41.69% compared to head (3986a3a) 40.35%. Report is 1 commits behind head on main.

Files	Patch %	Lines
ext/TITVisualizations.jl	0.00%	39 Missing :warning:
src/significance/basic_stat_significance.jl	0.00%	15 Missing :warning:
src/significance/surrogates_significance.jl	63.63%	8 Missing :warning:
src/visualizations.jl	0.00%	4 Missing :warning:
src/analysis/segmented_window.jl	0.00%	1 Missing :warning:
src/analysis/sliding_window.jl	95.00%	1 Missing :warning:
src/change_metrics/valuediff.jl	85.71%	1 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## main #66 +/- ## ========================================== - Coverage 41.69% 40.35% -1.35% ========================================== Files 17 19 +2 Lines 331 399 +68 ========================================== + Hits 138 161 +23 - Misses 193 238 +45 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.