feat: align v2 payloads with scrapegraph-py PR #82

Rework the MCP client/tool surface to produce payloads that exactly match
the v2 SDK contract (scrapegraph-py#82):

- scrape/crawl: emit `formats: [...]` entries (markdown/html/screenshot/
  branding/links/images/summary) instead of top-level format keys
- camelCase request keys everywhere (fetchConfig, numResults,
  locationGeoCode, maxDepth, maxPages, maxLinksPerPage, allowExternal,
  includePatterns, excludePatterns, contentTypes, webhookUrl, contentType)
- extract: use `schema` (not `output_schema`), support html/markdown body,
  default `mode: "normal"`, optional `contentType`
- searchscraper: add `locationGeoCode`, `timeRange`, `format`, `mode`,
  `prompt`, widen num_results range to 1-20
- smartcrawler_initiate: expose `max_links_per_page`, `allow_external`,
  `content_types`, widen formats beyond markdown/html
- monitor_create: wrap prompt+schema into a `{type: "json", ...}` format
  entry and expose `webhook_url`
- sgai_history: switch to page/limit/service, with an offset->page shim
- New tool: `generate_schema` backed by POST /schema

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: VinciGit00