perf: avoid heap allocation on .NET Core and .NET

Author: Flash0ver

src/Sentry/Internal/UnsampledTransaction.cs

@@ -12,7 +12,7 @@ internal sealed class UnsampledTransaction : NoOpTransaction
     // Although it's a little bit wasteful to create separate individual class instances here when all we're going to
     // report to sentry is the span count (in the client report), SDK users may refer to things like
     // `ITransaction.Spans.Count`, so we create an actual collection
-#if NETSTANDARD2_1_OR_GREATER
+#if NETCOREAPP2_0_OR_GREATER || NETSTANDARD2_1_OR_GREATER
     private readonly ConcurrentBag<ISpan> _spans = [];
 #else
     private ConcurrentBag<ISpan> _spans = [];
@@ -114,7 +114,7 @@ public ISpan StartChild(string operation, SpanId spanId)
 
     private void ReleaseSpans()
     {
-#if NETSTANDARD2_1_OR_GREATER
+#if NETCOREAPP2_0_OR_GREATER || NETSTANDARD2_1_OR_GREATER
         _spans.Clear();
 #else
         _spans = [];

src/Sentry/TransactionTracer.cs

@@ -168,7 +168,7 @@ public IReadOnlyList<string> Fingerprint
     /// <inheritdoc />
     public IReadOnlyDictionary<string, string> Tags => _tags;
 
-#if NETSTANDARD2_1_OR_GREATER
+#if NETCOREAPP2_0_OR_GREATER || NETSTANDARD2_1_OR_GREATER
     private readonly ConcurrentBag<ISpan> _spans = new();
 #else
     private ConcurrentBag<ISpan> _spans = new();
@@ -431,7 +431,7 @@ public string? Origin
 
     private void ReleaseSpans()
     {
-#if NETSTANDARD2_1_OR_GREATER
+#if NETCOREAPP2_0_OR_GREATER || NETSTANDARD2_1_OR_GREATER
         _spans.Clear();
 #else
         _spans = new ConcurrentBag<ISpan>();