Nested Schema doc parser confused about misleading labels

[t0yv0]

Our attempts to parse Nested Schema sections fail to make sense of some of the markdown. For example in dashboard.md there is a following section:

<<a id="nestedblock--widget--group_definition--widget--id--request--apm_query"></a>
### Nested Schema for `widget.group_definition.widget.id.request.show_present`

Required:

- **index** (String) Name of the index to query.

Optional:

- **compute** (Map of String, Deprecated) One of `compute` or `multi_compute` is required. The map has the keys as below. **Deprecated.** Define `compute_query` list with one element instead.
- **compute_query** (Block List, Max: 1) One of `compute_query` or `multi_compute` is required. The map has the keys as below. (see [below for nested schema](#nestedblock--widget--group_definition--widget--id--request--show_present--compute_query))
- **group_by** (Block List) Multiple `group_by` blocks are allowed with the structure below. (see [below for nested schema](#nestedblock--widget--group_definition--widget--id--request--show_present--group_by))
- **multi_compute** (Block List) One of `compute_query` or `multi_compute` is required. Multiple `multi_compute` blocks are allowed with the structure below. (see [below for nested schema](#nestedblock--widget--group_definition--widget--id--request--show_present--multi_compute))
- **search** (Map of String, Deprecated) Map defining the search query to use. **Deprecated.** Define `search_query` directly instead.
- **search_query** (String) The search query to use.

This could be an upstream issue. We could consider fixing upstream or injecting compensating data cleaning code.

Expected behavior

• a id=.. and Nested Schema for ... should point to the same property; in the example it seems that Nested Schema label is off

• The fully qualified property paths should be valid. In the example, widget.id.request subpath does not appear valid, as id is string-typed and should not have subproperties

Current behavior

Some nested schema sections have anomalies as above. The impact on Pulumi processing is mild, we fail to locate or misallocate property descriptions, but the actual code generation is not affected.

Steps to reproduce

Check out https://github.com/pulumi/pulumi-datadog/pull/126

Context (Environment)

Trying to upgrade pulumi-terraform-bridge to handle Nested Schema sections.

Affected feature

Documentation and docstrings on generated code.