docs: add NOTICE attribution per Apache-2.0 s4(d) (code AND weights)

NOTICE

@@ -0,0 +1,72 @@
+RTMPose-tiny — acaua mirror (pure-PyTorch port)
+=================================================
+
+This product includes:
+
+1. PORTED SOURCE CODE: a pure-PyTorch port of the RTMPose architecture
+   (located in the acaua repository at src/acaua/adapters/rtmpose/) is a
+   derivative work of:
+
+     - OpenMMLab's mmpose implementation
+       https://github.com/open-mmlab/mmpose @ commit
+       759b39c13fea6ba094afc1fa932f51dc1b11cbf9 — Apache-2.0
+       Files derived from:
+         mmpose/models/heads/coord_cls_heads/rtmcc_head.py  (RTMCCHead)
+         mmpose/models/utils/rtmcc_block.py                 (RTMCCBlock, Scale)
+         mmpose/models/utils/transformer.py                 (ScaleNorm)
+         mmpose/codecs/utils/post_processing.py             (get_simcc_maximum)
+         mmpose/codecs/simcc_label.py                       (SimCCLabel.decode)
+         mmpose/datasets/transforms/topdown_transforms.py   (TopdownAffine)
+         mmpose/structures/bbox/transforms.py               (get_warp_matrix,
+                                                            _rotate_point,
+                                                            _get_3rd_point)
+
+     - OpenMMLab's mmdetection implementation (CSPNeXt backbone)
+       https://github.com/open-mmlab/mmdetection @ commit
+       cfd5d3a985b0249de009b67d04f37263e11cdf3d — Apache-2.0
+       Files derived from:
+         mmdet/models/backbones/cspnext.py
+         mmdet/models/backbones/csp_darknet.py    (SPPBottleneck)
+         mmdet/models/layers/csp_layer.py         (CSPLayer, CSPNeXtBlock)
+         mmdet/models/layers/se_layer.py          (ChannelAttention)
+
+2. CONVERTED WEIGHTS: the model.safetensors file in this mirror is a
+   key-cleaned conversion of the upstream pretrained checkpoint:
+
+     - upstream URL:    https://download.openmmlab.com/mmpose/v1/projects/
+                        rtmposev1/rtmpose-tiny_simcc-aic-coco_pt-aic-coco_
+                        420e-256x192-cfc8f33d_20230126.pth
+     - upstream SHA256: e84eb5b9ee9432259bdd19d6a01156604ba27139ca6373ddb4
+                        ee7aa290d528e9
+     - upstream paper:  Jiang et al., "RTMPose: Real-Time Multi-Person Pose
+                        Estimation based on MMPose", arXiv:2303.07399
+     - training data:   AI Challenger (AIC) + COCO
+     - eval dataset:    COCO 17-keypoint (val 68.5 AP)
+
+   Conversion was performed by scripts/convert_rtmpose.py in the acaua
+   repository. The conversion only strips training-only buffers
+   (data_preprocessor.*); no parameter renames or reshapes are applied,
+   so the conversion is reversible and lossless.
+
+3. BUNDLED DETECTOR: the acaua RTMPoseAdapter at runtime composes with
+   CondadosAI/rtmdet_t_coco (a separate mirror with its own NOTICE and
+   provenance) to produce a top-down pose pipeline. The detector mirror
+   is a pure-PyTorch port of RTMDet-tiny under the same Apache-2.0
+   license; see https://huggingface.co/CondadosAI/rtmdet_t_coco/blob/main/
+   NOTICE for that mirror's full attribution chain.
+
+Mirrored on 2026-04-22 by CondadosAI.
+
+License
+-------
+Licensed under the Apache License, Version 2.0 (the "License"); you may
+not use this file except in compliance with the License. You may obtain
+a copy of the License at:
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.