fix #1962 Avoid extending Entry in Context1

[simonbasle]

The motivation for this change is that this drives two implementations
of Context#stream() to use the Context1 as an immutable map entry.

In turn, this proves problematic when the stream() is converted to a
string representation (eg. by loggers or assertions): the stream entries
use Context1{k=v} format, which is fine in a standalone toString but
less so in a representation of the whole stream.

AbstractMap.SimpleImmutableEntry can be leveraged instead, and it
has a proper key=value string representation by default.

[bsideup]

@simonbasle Context.putAll and ContextN.putAll use other.stream() which will create quite some allocations. Perhaps we should override them where possible and use the internal state

[simonbasle]

@bsideup default putAll is a bit wasteful for Context0-Context5, true: it invokes put multiple times (no real way around that if we want to check for duplicate keys and all) but also calls stream(), which unnecessarily creates map entries. This can be fixed by each specific implementation.

The default method should stay for compatibility, right? (since someone out there could have implemented Context without putAll)

I don't think the ContextN implementation can really be improved: it avoids intermediate new ContextX by preparing an internal map and feeding it both the current internal map and the other context as a map if it is not a ContextN itself. That last part limits the damages, because it implies there will be at most 5 "garbage" entries instantiated when creating the "other" map 👍

[simonbasle]

actually @bsideup there is no real way around the stream conversion, because the operation is not symmetric. The other values must be merged into the current context, potentially "replacing" existing keys. So we need some way of iterating, and the only one in the public API is stream()...

[codecov-io]

Codecov Report

Merging #1963 into master will increase coverage by 0.02%.
The diff coverage is 100%.

@@             Coverage Diff              @@
##             master    #1963      +/-   ##
============================================
+ Coverage     81.64%   81.66%   +0.02%     
+ Complexity     3982     3980       -2     
============================================
  Files           374      374              
  Lines         30811    30807       -4     
  Branches       5798     5798              
============================================
+ Hits          25156    25160       +4     
+ Misses         4060     4054       -6     
+ Partials       1595     1593       -2

Impacted Files	Coverage Δ	Complexity Δ
...e/src/main/java/reactor/util/context/Context1.java	100% <100%> (ø)	11 <1> (-3)	⬇️
...e/src/main/java/reactor/util/context/ContextN.java	100% <100%> (ø)	20 <1> (-1)	⬇️
.../java/reactor/core/scheduler/ElasticScheduler.java	84.07% <0%> (-1.77%)	26% <0%> (ø)
...r-core/src/main/java/reactor/core/Disposables.java	88.95% <0%> (-1.23%)	21% <0%> (ø)
.../java/reactor/core/publisher/BlockingIterable.java	78.12% <0%> (-1.05%)	7% <0%> (ø)
...ain/java/reactor/core/publisher/FluxPublishOn.java	86.57% <0%> (+0.2%)	6% <0%> (ø)	⬇️
...c/main/java/reactor/core/publisher/FluxExpand.java	95.51% <0%> (+0.44%)	3% <0%> (ø)	⬇️
...in/java/reactor/core/publisher/FluxWindowWhen.java	84.13% <0%> (+2.4%)	2% <0%> (ø)	⬇️
...va/reactor/core/publisher/ParallelMergeReduce.java	72.95% <0%> (+3.27%)	3% <0%> (ø)	⬇️
...ain/java/reactor/core/scheduler/SchedulerTask.java	76.78% <0%> (+3.57%)	17% <0%> (+1%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a7b11d8...4541db3. Read the comment docs.

[simonbasle]

@bsideup created #1964 to address your comment on limiting garbage generation. This is superseded by #1965

[simonbasle]

since there is a doubt about #1964, this PR is now no more a duplicate.