Raise error messages for invalid prompts

[jhj0517]

This PR raises the error message for invalid prompts.

As I understand it (please correct me if I'm wrong)
These prompts cannot be used:

1. Multiple box and points combinataions in SAM2ImagePredictor.predict() ( Only single combination is supported )

2. Multiple boxes in SAM2VideoPredictor.add_new_points_or_box() ( Only single box is supported )

This is sample code to reproduce errors with these invalid prompts:

import os
import numpy as np
from PIL import Image
from sam2.build_sam import build_sam2, build_sam2_video_predictor
from sam2.sam2_image_predictor import SAM2ImagePredictor

# Set paths
SAM2_CONFIGS_DIR = "path\\to\\configs_directory"
MODELS_DIR = "path\\to\\models_directory"
IMAGE_PATH = "path\\to\\image.jpg"
VIDEO_FRAMES_DIR = "path\\to\\video_frames_directory"

config_path = os.path.join(SAM2_CONFIGS_DIR, "sam2_hiera_l.yaml")
model_path = os.path.join(MODELS_DIR, "sam2_hiera_large.pt")

model = build_sam2(
    config_file=config_path,
    ckpt_path=model_path,
    device="cuda"
)
image_predictor = SAM2ImagePredictor(sam_model=model)
image = Image.open(IMAGE_PATH)
image_predictor.set_image(image)

double_box = np.array([
    [210, 609, 276, 698],
    [308, 425, 357, 501]
])
double_point_coords = np.array([
    [245, 655],
    [324, 456]
])
double_point_labels = np.array([1, 1])

# Using multiple combinations and getting errors
image_predictor.predict(point_coords=double_point_coords,
                        point_labels=double_point_labels,
                        box=double_box)

# Video Predictor case
model = build_sam2_video_predictor(
    config_file=config_path,
    ckpt_path=model_path,
    device="cuda"
)
inference_state = model.init_state(video_path=VIDEO_FRAMES_DIR)

# Using multiple boxes and getting errors
idx, scores, logits = model.add_new_points_or_box(
    frame_idx=0,
    obj_id=0,
    inference_state=inference_state,
    points=None,
    labels=None,
    box=double_box
)