docs: NumPy 1.x deprecation audit — plan, findings, and consolidated bug registry

Systematic comparison of NumSharp against NumPy 2.4.2 (v2.4.2) to identify
every feature corresponding to deprecated, removed, or behaviorally-changed
NumPy 1.x functionality.

numpy-1x-deprecation-audit.md (investigation plan):
- 7-phase audit covering removed APIs, type promotion (NEP 50), behavioral
  changes, dead code, documentation URLs, automated scripts, and
  NUMPY_NUMSHARP_MAP.md cross-check
- Grep results confirming which deprecated APIs are/aren't in NumSharp
- Python/C# scripts for automated verification

numpy-1x-deprecation-findings.md (results):
- 5 deprecated/removed functions still present (asscalar, find_common_type,
  round_, newbyteorder, asfarray reference)
- 13 deprecated aliases/constants in np.cs (NaN, Inf, float_, bool8, etc.)
- 12 type promotion mismatches in _typemap_arr_scalar (unsigned int array +
  signed int scalar — NEP 50 "array wins" rule)
- 6 behavioral divergences (floor/ceil int→double, unique return_inverse,
  asarray copy semantics, bool+bool, any(axis) broken, random legacy-only)
- 8 new bugs discovered (Bugs 27-63)
- 5 NUMPY_NUMSHARP_MAP.md corrections needed
- 385 stale documentation URLs across 121 files

OpenBugs.md (consolidated bug registry):
- All 54 bugs (7 fixed, 47 open) with full details
- Organized by category: dead code (13), wrong semantics (7), crashes (12),
  wrong values (15)
- Architectural root cause analysis (broadcast stride/data mismatch,
  multiple iteration paths, comparison operator inconsistency, dead code
  pattern)
- Dependency graph and recommended fix order for maximum cascade impact
- NumPy verification code for every bug

CLAUDE.md: Add GitHub Issues section, remove stale quick one-liners.

Author: Nucs

.claude/CLAUDE.md

@@ -306,6 +306,9 @@ switch (nd.typecode)
 }
 ```
 
+## GitHub Issues
+Create issues on `SciSharp/NumSharp` via `gh issue create` when the user requests it. `GH_TOKEN` is available in the environment.
+
 ## Build & Test
 
 ```bash
@@ -339,29 +342,6 @@ NumSharp uses unsafe in many places, hence include `#:property AllowUnsafeBlocks
 #:property AllowUnsafeBlocks=true
 ```
 
-### Quick One-Liners
-
-```bash
-# Run a script with full internal access
-dotnet run my_script.cs
-
-# Compare NumPy vs NumSharp type promotion
-dotnet run script.cs   # where script.cs contains:
-# var ct = np._FindCommonType(np.array(1), np.array(1.5));
-# Console.WriteLine(ct); // Double
-
-# Inspect Shape internals
-dotnet run script.cs   # where script.cs contains:
-# var s = new Shape(new int[]{2,3,4});
-# Console.WriteLine($"dims={string.Join(",",s.dimensions)} strides={string.Join(",",s.strides)} size={s.size}");
-
-# Check ViewInfo after slicing
-dotnet run script.cs   # where script.cs contains:
-# var arr = np.arange(24).reshape(2,3,4);
-# var sliced = arr["1, :, ::2"];
-# Console.WriteLine($"ViewInfo: {sliced.Shape.ViewInfo != null}, BroadcastInfo: {sliced.Shape.BroadcastInfo != null}");
-```
-
 ### Key Internal Members Available
 
 | Member | What it exposes |