Add reverse geocoding for brewery street addresses (#245)

* Add geocoding to seed application, remove postal codes from generator

Update comments

tooling/pipeline/includes/data_generation/generation_json_validation.h

@@ -3,9 +3,9 @@
 
 /**
  * @file data_generation/generation_json_validation.h
- * @brief Generator-agnostic helpers for preparing prompt context and
- * validating structured JSON output. Shared across all DataGenerator
- * implementations (Llama, OpenAI, ...).
+ * @brief Generator-agnostic helpers, shared across all DataGenerator
+ * implementations (Llama, OpenAI, ...), for preparing prompt context and
+ * validating structured JSON output.
  */
 
 #include <cstddef>
@@ -36,11 +36,9 @@ std::optional<std::string> ValidateBreweryJson(const std::string& raw,
                                                BreweryResult& brewery_out);
 
 /**
- * @brief Validates and parses user JSON output.
- *
- * Only populates `username`, `bio`, and `activity_weight` -- `first_name`
- * and `last_name` are not LLM-authored and are set separately from the
- * sampled Name.
+ * @brief Validates and parses user JSON output, populating only
+ * `username`, `bio`, and `activity_weight` (`first_name` and `last_name`
+ * come from the sampled Name).
  *
  * @param raw Raw model output.
  * @param user_out Parsed user payload.

tooling/pipeline/includes/data_generation/openai_generator.h

@@ -18,13 +18,9 @@
 
 /**
  * @brief Data generator implementation backed by the OpenAI Chat
- * Completions API.
- *
- * Unlike LlamaGenerator, this generator has no local model state and no
- * chat-template formatting step -- Chat Completions takes system/user
- * messages directly, and Structured Outputs (response_format: json_schema,
- * strict: true) guarantee schema-valid JSON, replacing llama.cpp's GBNF
- * grammar constraint.
+ * Completions API, sending system/user messages directly and using
+ * Structured Outputs (response_format: json_schema, strict: true) to
+ * guarantee schema-valid JSON.
  */
 class OpenAIGenerator final : public DataGenerator {
   public:

tooling/pipeline/includes/data_generation/openai_json_schemas.h

@@ -4,14 +4,9 @@
 /**
  * @file data_generation/openai_json_schemas.h
  * @brief JSON Schemas constraining structured output from OpenAIGenerator
- * Chat Completions requests, via response_format.json_schema (Structured
- * Outputs, strict mode). This is OpenAI's equivalent of the GBNF grammars
- * in json_grammars.h -- the API guarantees the response message content is
- * a single JSON string that validates against the schema, so there's no
- * grammar-level "thought-block" prefix to account for.
- *
- * OpenAI's strict mode requires "additionalProperties": false on every
- * object in the schema, which these already have.
+ * Chat Completions requests via response_format.json_schema (Structured
+ * Outputs, strict mode), each setting "additionalProperties": false as
+ * strict mode requires.
  */
 
 #include <string_view>