[MRESOLVER-324] Make bf option as default (#325)

Make new BF resolver the default.

---

https://issues.apache.org/jira/browse/MRESOLVER-324

maven-resolver-impl/src/main/java/org/eclipse/aether/internal/impl/collect/DefaultDependencyCollector.java

@@ -46,7 +46,7 @@
 public class DefaultDependencyCollector implements DependencyCollector, Service {
     private static final String CONFIG_PROP_COLLECTOR_IMPL = "aether.dependencyCollector.impl";
 
-    private static final String DEFAULT_COLLECTOR_IMPL = DfDependencyCollector.NAME;
+    private static final String DEFAULT_COLLECTOR_IMPL = BfDependencyCollector.NAME;
 
     private final Map<String, DependencyCollectorDelegate> delegates;
 

src/site/markdown/configuration.md

@@ -64,7 +64,7 @@ Option | Type | Description | Default Value | Supports Repo ID Suffix
 `aether.connector.wagon.config` | Object | The configuration to use for the Wagon provider. | - | yes (must be used)
 `aether.dependencyCollector.maxCycles` | int | Only up to the given amount cyclic dependencies are emitted. | `10` | no
 `aether.dependencyCollector.maxExceptions` | int | Only exceptions up to the number given in this configuration property are emitted. Exceptions which exceed that number are swallowed. | `50` | no
-`aether.dependencyCollector.impl` | String | The name of the dependency collector implementation to use: depth-first (original) named `df`, and breadth-first (new in 1.8.0) named `bf`. Both collectors produce equivalent results, but they may differ performance wise, depending on project being applied to. Our experience shows that existing `df` is well suited for smaller to medium size projects, while `bf` may perform better on huge projects with many dependencies. Experiment (and come back to us!) to figure out which one suits you the better. | `"df"` | no
+`aether.dependencyCollector.impl` | String | The name of the dependency collector implementation to use: depth-first (original) named `df`, and breadth-first (new in 1.8.0) named `bf`. Both collectors produce equivalent results, but they may differ performance wise, depending on project being applied to. Our experience shows that existing `df` is well suited for smaller to medium size projects, while `bf` may perform better on huge projects with many dependencies. Experiment (and come back to us!) to figure out which one suits you the better. | `"bf"` | no
 `aether.dependencyCollector.bf.skipper` | boolean | Flag controlling whether to skip resolving duplicate/conflicting nodes during the breadth-first (`bf`) dependency collection process. | `true` | no
 `aether.dependencyCollector.bf.threads` or `maven.artifact.threads` | int | Number of threads to use for collecting POMs and version ranges in BF collector. | `5` | no
 `aether.dependencyCollector.pool.artifact` | String | Flag controlling interning data pool type used by dependency collector for Artifact instances, matters for heap consumption. By default uses "weak" references (consume less heap). Using "hard" will make it much more memory aggressive and possibly faster (system and Java dependent). Supported values: `"hard"`, `"weak"`. | `"weak"` | no