feat: first-class data constraints (and ranges)

**BEFORE**:
```python
class Experiment:
temperature: float in (0, 100) # °C
ph: float in [0, 14]
concentration: float > 0
sample_size: int % 2 == 0 # Must be even
```

**AFTER**:
```python
# After (transpiled.py)
from vyper.validation import Range, InRange, validate_unit
from pydantic import BaseModel, validator
from typing import Any

class ChemicalReaction(BaseModel):
temperature: float
concentration: float
catalyst_amount: float

@validator('temperature')
def validate_temperature(cls, v):
if not 0 < v < 100:
raise ValueError(f"Temperature {v}°C not in range (0,100)")
return v

@validator('concentration')
def validate_concentration(cls, v):
if not v > 0:
raise ValueError("Concentration must be positive")
validate_unit(v, 'mg/L')
return v

@validator('catalyst_amount')
def validate_catalyst(cls, v):
if not 0.1 <= v <= 1.0:
raise ValueError("Catalyst amount must be between 0.1 and 1.0 mg")
validate_unit(v, 'mg')
return v
```

The issue names no source files, tests, or entry points; start by tracing how annotated class fields are parsed and transpiled to the shown Python output. Define the supported constraint syntax and validation behavior, then add tests covering ranges, comparisons, units, and errors.

Written by the indexing model from the issue text.