[Docs] Fully qualify standard library imports with std. prefix

Summary
As of Mojo 1.0.0b2, implicit standard-library imports are a hard error —
stdlib modules must be fully qualified with a std. prefix (e.g. from
std.gpu.host import DeviceContext instead of from gpu.host import
DeviceContext). This was previously a deprecation warning in b1. Several
doc/tutorial code examples still used the old syntax, so readers copying
them now hit an unhelpful unable to locate module 'gpu' error.

This PR migrates the remaining stale examples across the docs to the
std.-qualified form. Reported in [forum
#3237](https://forum.modular.com/t/standard-library-import-warning-error-between-b1-and-b2/3237).

MODULAR_ORIG_COMMIT_REV_ID: 8ddba94a5b0a5b29b29f2baa89bf340227b01a79

Author: dunnoconnor

book/i18n/ko/src/puzzle_23/benchmarking.md

@@ -103,7 +103,7 @@ Benchmarks completed!
 벤치마킹 시스템은 Mojo의 내장 `benchmark` 모듈을 사용합니다:
 
 ```mojo
-from benchmark import Bench, BenchConfig, Bencher, BenchId, keep
+from std.benchmark import Bench, BenchConfig, Bencher, BenchId, keep
 bench_config = BenchConfig(max_iters=10, num_warmup_iters=1)
 ```
 

book/i18n/ko/src/puzzle_24/warp_extra.md

@@ -52,12 +52,12 @@
 
 ```mojo
 # 리덕션 연산
-from gpu.primitives.warp import sum, max
+from std.gpu.primitives.warp import sum, max
 var total = sum(partial_values)
 var maximum = max(partial_values)
 
 # 통신 패턴
-from gpu.primitives.warp import shuffle_idx, prefix_sum
+from std.gpu.primitives.warp import shuffle_idx, prefix_sum
 var broadcast = shuffle_idx(my_value, 0)
 var running_sum = prefix_sum(my_value)
 ```

book/i18n/ko/src/puzzle_24/warp_simt.md

@@ -13,7 +13,7 @@
 **간단한 예시:**
 
 ```mojo
-from gpu.primitives.warp import sum
+from std.gpu.primitives.warp import sum
 # 워프 내 32개 스레드가 동시에 실행:
 var my_value = input[my_thread_id]     # 각 스레드가 서로 다른 데이터를 가져옴
 var warp_total = sum(my_value)         # 모든 스레드가 하나의 합계에 기여
@@ -56,7 +56,7 @@ var result = a + b # 8쌍을 동시에 덧셈
 
 ```mojo
 # 스레드 기반 코드가 벡터 연산으로 변환됩니다
-from gpu.primitives.warp import sum
+from std.gpu.primitives.warp import sum
 
 var my_data = input[thread_id]         # 각 스레드가 자기 요소를 가져옴
 var partial = my_data * coefficient    # 모든 스레드가 동시에 계산
@@ -169,8 +169,8 @@ GPU는 지연 시간이 아닌 **처리량**에 최적화되어 있습니다. SI
 워프 내 각 스레드는 0부터 `WARP_SIZE-1`까지의 **레인 ID**를 갖습니다:
 
 ```mojo
-from gpu import lane_id
-from gpu.primitives.warp import WARP_SIZE
+from std.gpu import lane_id
+from std.gpu.primitives.warp import WARP_SIZE
 
 # 커널 함수 내에서:
 my_lane = lane_id()  # 0-31 (NVIDIA/RDNA) 또는 0-63 (CDNA) 반환
@@ -192,7 +192,7 @@ barrier()  # 명시적 동기화 필요
 var total = shared[0] + shared[1] + ... + shared[WARP_SIZE] # 합산 리덕션
 
 # 2. 워프 방식:
-from gpu.primitives.warp import sum
+from std.gpu.primitives.warp import sum
 
 var total = sum(partial_result)  # 암시적 동기화!
 ```
@@ -306,7 +306,7 @@ else:
 ### NVIDIA vs AMD 워프 크기
 
 ```mojo
-from gpu.primitives.warp import WARP_SIZE
+from std.gpu.primitives.warp import WARP_SIZE
 
 # NVIDIA GPUs:     WARP_SIZE = 32
 # AMD RDNA GPUs:   WARP_SIZE = 32 (wavefront32 모드)

book/i18n/ko/src/puzzle_24/warp_sum.md

@@ -324,8 +324,8 @@ if lane_id() == 0:
 ### 4. **import에서 사용 가능한 함수들**
 
 ```mojo
-from gpu import lane_id
-from gpu.primitives.warp import sum as warp_sum, WARP_SIZE
+from std.gpu import lane_id
+from std.gpu.primitives.warp import sum as warp_sum, WARP_SIZE
 
 # 함수 내에서:
 my_lane = lane_id()           # 0 ~ WARP_SIZE-1

book/src/puzzle_23/benchmarking.md

@@ -102,7 +102,7 @@ Benchmarks completed!
 The benchmarking system uses Mojo's built-in `benchmark` module:
 
 ```mojo
-from benchmark import Bench, BenchConfig, Bencher, BenchId, keep
+from std.benchmark import Bench, BenchConfig, Bencher, BenchId, keep
 bench_config = BenchConfig(max_iters=10, num_warmup_iters=1)
 ```
 

book/src/puzzle_24/warp_extra.md

@@ -50,12 +50,12 @@
 
 ```mojo
 # Reduction operations
-from gpu.primitives.warp import sum, max
+from std.gpu.primitives.warp import sum, max
 var total = sum(partial_values)
 var maximum = max(partial_values)
 
 # Communication patterns
-from gpu.primitives.warp import shuffle_idx, prefix_sum
+from std.gpu.primitives.warp import shuffle_idx, prefix_sum
 var broadcast = shuffle_idx(my_value, 0)
 var running_sum = prefix_sum(my_value)
 ```

book/src/puzzle_24/warp_simt.md

@@ -12,7 +12,7 @@ processor.
 **Simple example:**
 
 ```mojo
-from gpu.primitives.warp import sum
+from std.gpu.primitives.warp import sum
 # All 32 threads in the warp execute this simultaneously:
 var my_value = input[my_thread_id]     # Each gets different data
 var warp_total = sum(my_value)         # All contribute to one sum
@@ -55,7 +55,7 @@ var result = a + b # Add 8 pairs simultaneously
 
 ```mojo
 # Thread-based code that becomes vector operations
-from gpu.primitives.warp import sum
+from std.gpu.primitives.warp import sum
 
 var my_data = input[thread_id]         # Each thread gets its element
 var partial = my_data * coefficient    # All threads compute simultaneously
@@ -168,8 +168,8 @@ GPUs are optimized for **throughput**, not latency. SIMT enables:
 Each thread within a warp has a **lane ID** from 0 to `WARP_SIZE-1`:
 
 ```mojo
-from gpu import lane_id
-from gpu.primitives.warp import WARP_SIZE
+from std.gpu import lane_id
+from std.gpu.primitives.warp import WARP_SIZE
 
 # Within a kernel function:
 my_lane = lane_id()  # Returns 0-31 (NVIDIA/RDNA) or 0-63 (CDNA)
@@ -191,7 +191,7 @@ barrier()  # Explicit synchronization required
 var total = shared[0] + shared[1] + ... + shared[WARP_SIZE] # Sum reduction
 
 # 2. Warp approach:
-from gpu.primitives.warp import sum
+from std.gpu.primitives.warp import sum
 
 var total = sum(partial_result)  # Implicit synchronization!
 ```
@@ -307,7 +307,7 @@ else:
 ### NVIDIA vs AMD warp sizes
 
 ```mojo
-from gpu.primitives.warp import WARP_SIZE
+from std.gpu.primitives.warp import WARP_SIZE
 
 # NVIDIA GPUs:     WARP_SIZE = 32
 # AMD RDNA GPUs:   WARP_SIZE = 32 (wavefront32 mode)

book/src/puzzle_24/warp_sum.md

@@ -328,8 +328,8 @@ functional approach.
 ### 4. **Available functions from imports**
 
 ```mojo
-from gpu import lane_id
-from gpu.primitives.warp import sum as warp_sum, WARP_SIZE
+from std.gpu import lane_id
+from std.gpu.primitives.warp import sum as warp_sum, WARP_SIZE
 
 # Inside your function:
 my_lane = lane_id()           # 0 to WARP_SIZE-1